phi-4-mm debug 1

=== Critical Debug Info ===
1. Top-level modules: ['base_model']
2. All modules containing 'audio': ['base_model.model.base_model.model.model.embed_tokens_extend.audio_embed', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.drop', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.embed', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.embed.out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.embed.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.embed.conv.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.embed.conv.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.embed.conv.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.embed.conv.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.embed.conv.5', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.embed.conv.6', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.pos_emb', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.pos_emb.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.relative_attention_bias_layer', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.relative_attention_bias_layer.bias_values', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.0._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.1._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.2._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.3._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.4._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.5._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.6._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.7._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.8._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.9._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.10._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.11._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.12._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.13._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.14._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.15._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.16._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.17._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.18._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.19._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.20._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.21._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.22._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_in.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn.linear_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn.linear_k', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn.linear_v', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn.linear_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn.quant_q', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn.quant_x', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn.dequant', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn.ffunc', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.self_attn.ffunc.activation_post_process', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.ln1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.glu', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.glu.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.glu.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.glu.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.ext_pw_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.squeeze_excitation', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.dw_sep_conv_1d', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.dw_sep_conv_1d.dw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.conv.dw_sep_conv_1d.pw_conv', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out.net', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out.net.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out.net.0.linear', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out.net.0.glu_act.act_fn.act_fn', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out.net.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out.net.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.feed_forward_out.net.3', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.layer_norm_att', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoders.23._checkpoint_wrapped_module.layer_norm', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.encoder.encoder_embedding', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.base_layer', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_dropout.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_A.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_B.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_embedding_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_embedding_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_magnitude_vector', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.base_layer', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_dropout.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_A.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_B.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_embedding_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_embedding_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_magnitude_vector', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.vision', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.vision.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.vision.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.vision.2']
3. All modules containing 'speech': ['base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.base_layer', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_dropout.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_A.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_B.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_embedding_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_embedding_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_magnitude_vector', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.base_layer', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_dropout.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_A.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_B.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_embedding_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_embedding_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_B.speech']
4. Projection layers: ['base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.0.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.0.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.0.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.0.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.1.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.1.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.1.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.1.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.2.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.2.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.2.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.2.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.3.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.3.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.3.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.3.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.4.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.4.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.4.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.4.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.5.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.5.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.5.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.5.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.6.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.6.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.6.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.6.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.7.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.7.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.7.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.7.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.8.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.8.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.8.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.8.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.9.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.9.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.9.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.9.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.10.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.10.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.10.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.10.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.11.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.11.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.11.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.11.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.12.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.12.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.12.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.12.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.13.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.13.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.13.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.13.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.14.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.14.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.14.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.14.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.15.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.15.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.15.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.15.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.16.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.16.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.16.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.16.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.17.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.17.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.17.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.17.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.18.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.18.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.18.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.18.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.19.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.19.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.19.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.19.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.20.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.20.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.20.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.20.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.21.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.21.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.21.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.21.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.22.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.22.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.22.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.22.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.23.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.23.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.23.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.23.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.24.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.24.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.24.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.24.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.25.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.25.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.25.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.25.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.26.self_attn.k_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.26.self_attn.v_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.26.self_attn.q_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.encoder.layers.26.self_attn.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_processor.head.attention.out_proj', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_projection', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_projection.0', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_projection.1', 'base_model.model.base_model.model.model.embed_tokens_extend.image_embed.img_projection.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.base_layer', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_dropout.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_A.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_B.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_embedding_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_embedding_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.0.lora_magnitude_vector', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.base_layer', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_dropout', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_dropout.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_A.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_B.default', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_embedding_A', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_embedding_B', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.speech.2.lora_magnitude_vector', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.vision', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.vision.0', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.vision.1', 'base_model.model.base_model.model.model.embed_tokens_extend.audio_embed.audio_projection.vision.2', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.0.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.0.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.0.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.0.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.1.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.1.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.1.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.1.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.2.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.2.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.2.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.2.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.3.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.3.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.3.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.3.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.4.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.4.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.4.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.4.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.5.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.5.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.5.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.5.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.6.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.6.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.6.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.6.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.7.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.7.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.7.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.7.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.8.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.8.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.8.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.8.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.9.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.9.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.9.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.9.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.10.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.10.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.10.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.10.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.11.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.11.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.11.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.11.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.12.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.12.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.12.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.12.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.13.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.13.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.13.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.13.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.14.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.14.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.14.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.14.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.15.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.15.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.15.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.15.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.16.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.16.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.16.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.16.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.17.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.17.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.17.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.17.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.18.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.18.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.18.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.18.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.19.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.19.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.19.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.19.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.20.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.20.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.20.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.20.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.21.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.21.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.21.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.21.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.22.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.22.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.22.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.22.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.23.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.23.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.23.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.23.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.24.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.24.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.24.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.24.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.25.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.25.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.25.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.25.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.26.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.26.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.26.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.26.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.27.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.27.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.27.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.27.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.28.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.28.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.28.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.28.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.29.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.29.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.29.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.29.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.30.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.30.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.30.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.30.mlp.down_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.base_layer', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_A', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_B', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.31.self_attn.o_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.base_layer', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_A', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_B', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.31.self_attn.qkv_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.base_layer', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_A', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_B', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.31.mlp.gate_up_proj.lora_magnitude_vector', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.base_layer', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_dropout', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_dropout.vision', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_dropout.speech', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_A', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_A.vision', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_A.speech', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_B', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_B.vision', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_B.speech', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_embedding_A', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_embedding_B', 'base_model.model.base_model.model.model.layers.31.mlp.down_proj.lora_magnitude_vector']
Total params: 5,574,787,904 | Trainable: 327,680