Hi, I am trying using my “static_quantize_weights.pt” to do QAT(or you can call finetuning)；
But it come across the problem that quantized-weight doesn’t have the same key compare with the original model.
the ERROR message is like below:
##########
Traceback (most recent call last):
File “Static_quantize5.py”, line 200, in
train(hyp, opt = opt,device = torch.device(‘cpu’), tb_writer = tb_writer)
File “/home/gene/YuRen/yolov7/train.py”, line 102, in train
state_dict = ckpt[‘model’].float().state_dict() # to FP32
KeyError: ‘model’
############
How could I do next?
Note: Pytorch 2.1.0，FX-graph-Mode Post-training-static-quantization with YOLOv7
Please help, Thank you!!!
Here are the key from original and from static-quantized-weights:
Original keys:
epoch
best_fitness
training_results
model
ema
updates
optimizer
wandb_id
static-quantized-weight 's keys: (Too many, Just a part of keys)
model_0_conv_input_scale_0
model_0_conv_input_zero_point_0
model_39_scale_0
model_39_zero_point_0
model_49_scale_0
model_49_zero_point_0
_tensor_constant0
_tensor_constant1
_tensor_constant2
_tensor_constant3
_tensor_constant4
model_77_scale_4
model_77_zero_point_4
_tensor_constant5
_tensor_constant6
model_77_scale_10
model_77_zero_point_10
_tensor_constant7
_tensor_constant8
model_77_scale_16
model_77_zero_point_16
model.0.conv.weight
model.0.conv.bias
model.0.conv.scale
model.0.conv.zero_point
model.0.act.scale
model.0.act.zero_point
model.1.conv.weight
model.1.conv.bias
model.1.conv.scale
model.1.conv.zero_point
model.1.act.scale
model.1.act.zero_point
model.2.conv.weight
model.2.conv.bias
model.2.conv.scale
model.2.conv.zero_point
model.2.act.scale
model.2.act.zero_point
model.3.conv.weight
model.3.conv.bias
model.3.conv.scale
model.3.conv.zero_point
model.3.act.scale
model.3.act.zero_point
model.4.conv.weight
model.4.conv.bias
model.4.conv.scale
model.4.conv.zero_point
model.4.act.scale
model.4.act.zero_point
model.5.conv.weight
model.5.conv.bias
model.5.conv.scale
model.5.conv.zero_point