Axolotl training on private data

axolotl.yaml

base_model: google/gemma-3-27b-it
#model_type: AutoModelForCausalLM
#tokenizer_type: AutoTokenizer

deepspeed: deepspeed_config.json

load_in_8bit: false
load_in_4bit: true
#strict: false

rl: orpo
orpo_alpha: 0.1


chat_template: gemma
datasets:
  - path: mlabonne/orpo-dpo-mix-40k
    type: chat_template.argilla











dataset_prepared_path: last_run_prepared
val_set_size: 0.1
output_dir: ./orpo-lora-out

adapter: qlora

sequence_len: 4096
sample_packing: false
pad_to_sequence_len: true

save_safetensors: true

lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
#lora_target_linear: true
lora_fan_in_fan_out:
#lora_target_modules:
#  - gate_proj
#  - down_proj
#  - up_proj
#  - q_proj
#  - v_proj
#  - k_proj
#  - o_proj

gradient_accumulation_steps: 1
micro_batch_size: 2
num_epochs: 1
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.00000800

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
auto_resume_from_checkpoints: true
local_rank:
logging_steps: 100
save_steps: .25
xformers_attention:
flash_attention: true


warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
debug:
weight_decay: 0.0
fsdp:
fsdp_config:

special_tokens:
  bos_token: "<bos>"
  eos_token: "<eos>"
  unk_token: "<unk>"


lora_model_dir: /path/to/lora/adapters