leslie-fang-intel

import torch
import torch._dynamo as torchdynamo
from torch.ao.quantization import (
    get_default_qconfig,
    QConfigMapping,
)
from torch.ao.quantization._quantize_pt2e import prepare_pt2e, convert_pt2e

class Mod(torch.nn.Module):
    def __init__(self,) -> None:

leslie-fang-intel

import torch
import torch._dynamo as torchdynamo
from torch.ao.quantization import (
    get_default_qconfig,
    QConfigMapping,
)
from torch.ao.quantization._quantize_pt2e import prepare_pt2e, convert_pt2e

class Mod(torch.nn.Module):
    def __init__(self,) -> None:

leslie-fang-intel

import torch
from torch import _dynamo, _inductor
from torch._inductor import config
import logging
import numpy as np
import random
from torch._inductor import codecache, config, metrics, test_operators
import torch.ao.quantization.fx._decomposed

torch._dynamo.config.verbose = True

leslie-fang-intel

import torch
import torchvision
import torch._dynamo as torchdynamo
import copy
from torch.ao.quantization._pt2e.quantizer import (
    QNNPackQuantizer,
)
from torch.ao.quantization._quantize_pt2e import (
    convert_pt2e,
    prepare_qat_pt2e_quantizer,

leslie-fang-intel

Error Message:
╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮
│ /home/lesliefang/pytorch_1_7_1/torch_inductor/torch_script/inductor/int8/test_dynamo_export_none │
│ _full_graph/test_Llama_dynamo_export.py:43 in <module>                                           │
│                                                                                                  │
│   40 │   return                                                                                  │
│   41                                                                                             │
│   42 if __name__ == "__main__":                                                                  │
│ ❱ 43 │   test_Llama()                                                                            │
│   44                                                                                             │

leslie-fang-intel

graph(%self : __torch__.torch.fx.graph_module.___torch_mangle_1.GraphModule,
      %x : Tensor):
  %self._param_constant160 : Float(1000, strides=[1], requires_grad=0, device=cpu) = prim::Constant[value=<Tensor>]()
  %self._param_constant156_bias : Float(2048, strides=[1], requires_grad=0, device=cpu) = prim::Constant[value=<Tensor>]()
  %self._param_constant153_bias : Float(512, strides=[1], requires_grad=0, device=cpu) = prim::Constant[value=<Tensor>]()
  %self._param_constant150_bias : Float(512, strides=[1], requires_grad=0, device=cpu) = prim::Constant[value=<Tensor>]()
  %self._param_constant147_bias : Float(2048, strides=[1], requires_grad=0, device=cpu) = prim::Constant[value=<Tensor>]()
  %self._param_constant144_bias : Float(512, strides=[1], requires_grad=0, device=cpu) = prim::Constant[value=<Tensor>]()
  %self._param_constant141_bias : Float(512, strides=[1], requires_grad=0, device=cpu) = prim::Constant[value=<Tensor>]()
  %self._param_constant138_bias : Float(2048, strides=[1], requires_grad=0,

leslie-fang-intel

graph(%self.1 : __torch__.intel_extension_for_pytorch.quantization._quantize_utils.___torch_mangle_194.QuantizationDispatchModule,
      %x : Tensor):
  %2 : float = prim::Constant[value=0.020705882459878922]() # /home/lesliefang/pytorch_1_7_1/quantization/frameworks.ai.pytorch.private-cpu/torch/_tensor.py:1298:0
  %3 : int = prim::Constant[value=0]() # /home/lesliefang/pytorch_1_7_1/quantization/frameworks.ai.pytorch.private-cpu/torch/_tensor.py:1298:0
  %4 : int = prim::Constant[value=12]() # /home/lesliefang/pytorch_1_7_1/quantization/frameworks.ai.pytorch.private-cpu/torch/_tensor.py:1298:0
  %5 : int = prim::Constant[value=1]() # /home/lesliefang/pytorch_1_7_1/quantization/frameworks.ai.pytorch.private-cpu/torch/_tensor.py:1298:0
  %6 : int[] = prim::Constant[value=[1, 1]]()
  %7 : int = prim::Constant[value=-1]() # /home/lesliefang/pytorch_1_7_1/quantization/frameworks.ai.pytorch.private-cpu/torch/_tensor.py:1298:0
  %ret.1 : Tensor = aten::quantize_per_tensor(%x, %2, %3, %4) # /home/lesliefang/pytorch_

leslie-fang-intel

def forward(self, x):
    arg0, = fx_pytree.tree_flatten_spec(([x], {}), self._in_spec)
    _scale_0 = self._scale_0
    _zero_point_0 = self._zero_point_0
    quantize_per_tensor = torch.ops.quantized_decomposed.quantize_per_tensor(arg0, _scale_0, _zero_point_0, 0, 255, torch.uint8);  arg0 = None
    dequantize_per_tensor = torch.ops.quantized_decomposed.dequantize_per_tensor(quantize_per_tensor, _scale_0, _zero_point_0, 0, 255, torch.uint8);  quantize_per_tensor = _scale_0 = _zero_point_0 = None
    _param_constant0 = self._param_constant0
    conv_scale_0 = self.conv_scale_0
    conv_zero_point_0 = self.conv_zero_point_0
    quantize_per_channel = torch.ops.quantized_decomposed.quantize_per_channel(_param_constant0, conv_scale_0, conv_zero_point_0, 0, -128, 127, torch.int8);  _param_constant0 = None

leslie-fang-intel

import torch
import torch.nn as nn
import torch._dynamo as torchdynamo
import copy

from torch.ao.quantization._quantize_pt2e import (
    convert_pt2e,
)
from torch._inductor.compile_fx import compile_fx
import torch.ao.quantization._pt2e.quantizer.qnnpack_quantizer as qq

leslie-fang-intel

import copy
import itertools
import operator
from typing import Callable, Dict, List, Optional, Set, Any

import torch
import torch._dynamo as torchdynamo
from torch.ao.quantization._pt2e.quantizer.utils import (
    _annotate_input_qspec_map,
    _annotate_output_qspec,