我想通过在变压器中的
compute losstrainer.py我尝试添加
print logger.infodef compute_loss(self, model, inputs, return_outputs=False):
"""
How the loss is computed by Trainer. By default, all models return the loss in the first element.
Subclass and override for custom behavior.
"""
if self.label_smoother is not None and "labels" in inputs:
labels = inputs.pop("labels")
else:
labels = None
logger.info(f"{labels}") # here~!
outputs = model(**inputs)
# Save past state if it exists
# TODO: this needs to be fixed and made cleaner later.
if self.args.past_index >= 0:
self._past = outputs[self.args.past_index]
if labels is not None:
if is_peft_available() and isinstance(model, PeftModel):
model_name = unwrap_model(model.base_model)._get_name()
else:
model_name = unwrap_model(model)._get_name()
if model_name in MODEL_FOR_CAUSAL_LM_MAPPING_NAMES.values():
loss = self.label_smoother(outputs, labels, shift_labels=True)
else:
loss = self.label_smoother(outputs, labels)
else:
if isinstance(outputs, dict) and "loss" not in outputs:
raise ValueError(
"The model did not return a loss from the inputs, only the following keys: "
f"{','.join(outputs.keys())}. For reference, the inputs it received are {','.join(inputs.keys())}."
)
# We don't use .loss here since the model may return tuples instead of ModelOutput.
loss = outputs["loss"] if isinstance(outputs, dict) else outputs[0]
return (loss, outputs) if return_outputs else loss
终端显示：
(qwen) seamoon@gpu07:~/code/Qwen$ bash finetune/finetune_ds.sh
[2024-01-06 17:55:53,192] torch.distributed.run: [WARNING]
[2024-01-06 17:55:53,192] torch.distributed.run: [WARNING] *****************************************
[2024-01-06 17:55:53,192] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
[2024-01-06 17:55:53,192] torch.distributed.run: [WARNING] *****************************************
[2024-01-06 17:55:54,784] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2024-01-06 17:55:54,853] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)
The model is automatically converting to bf16 for faster inference. If you want to disable the automatic precision, please manually add bf16/fp16/fp32=True to "AutoModelForCausalLM.from_pretrained".
Try importing flash-attention for faster inference...
Warning: import flash_attn rotary fail, please install FlashAttention rotary to get higher efficiency https://github.com/Dao-AILab/flash-attention/tree/main/csrc/rotary
Warning: import flash_attn rms_norm fail, please install FlashAttention layer_norm to get higher efficiency https://github.com/Dao-AILab/flash-attention/tree/main/csrc/layer_norm
Warning: import flash_attn fail, please install FlashAttention to get higher efficiency https://github.com/Dao-AILab/flash-attention
The model is automatically converting to bf16 for faster inference. If you want to disable the automatic precision, please manually add bf16/fp16/fp32=True to "AutoModelForCausalLM.from_pretrained".
Try importing flash-attention for faster inference...
Warning: import flash_attn rotary fail, please install FlashAttention rotary to get higher efficiency https://github.com/Dao-AILab/flash-attention/tree/main/csrc/rotary
Warning: import flash_attn rms_norm fail, please install FlashAttention layer_norm to get higher efficiency https://github.com/Dao-AILab/flash-attention/tree/main/csrc/layer_norm
Warning: import flash_attn fail, please install FlashAttention to get higher efficiency https://github.com/Dao-AILab/flash-attention
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 2.41it/s]
Loading data...
Formatting inputs...Skip in lazy mode
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:01<00:00, 1.10it/s]
0%| | 0/20 [00:00<?, ?it/s]/home/seamoon/anaconda3/envs/qwen/lib/python3.10/site-packages/torch/utils/checkpoint.py:429: UserWarning: torch.utils.checkpoint: please pass in use_reentrant=True or use_reentrant=False explicitly. The default value of use_reentrant will be updated to be False in the future. To maintain current behavior, pass use_reentrant=True. It is recommended that you use use_reentrant=False. Refer to docs for more details on the differences between the two variants.
warnings.warn(
/home/seamoon/anaconda3/envs/qwen/lib/python3.10/site-packages/torch/utils/checkpoint.py:429: UserWarning: torch.utils.checkpoint: please pass in use_reentrant=True or use_reentrant=False explicitly. The default value of use_reentrant will be updated to be False in the future. To maintain current behavior, pass use_reentrant=True. It is recommended that you use use_reentrant=False. Refer to docs for more details on the differences between the two variants.
warnings.warn(
{'loss': 1.1746, 'learning_rate': 1e-05, 'epoch': 0.24}
{'loss': 1.2259, 'learning_rate': 9.931806517013612e-06, 'epoch': 0.48}
{'loss': 0.1804, 'learning_rate': 9.729086208503174e-06, 'epoch': 0.73}
{'loss': 0.1697, 'learning_rate': 9.397368756032445e-06, 'epoch': 0.97}
{'loss': 0.1526, 'learning_rate': 8.94570254698197e-06, 'epoch': 1.21}
{'loss': 0.1521, 'learning_rate': 8.386407858128707e-06, 'epoch': 1.45}
{'loss': 0.1554, 'learning_rate': 7.734740790612137e-06, 'epoch': 1.7}
{'loss': 0.1513, 'learning_rate': 7.008477123264849e-06, 'epoch': 1.94}
{'loss': 0.1523, 'learning_rate': 6.227427435703997e-06, 'epoch': 2.18}
{'loss': 0.1326, 'learning_rate': 5.412896727361663e-06, 'epoch': 2.42}
{'loss': 0.1484, 'learning_rate': 4.587103272638339e-06, 'epoch': 2.67}
{'loss': 0.1335, 'learning_rate': 3.7725725642960047e-06, 'epoch': 2.91}
{'loss': 0.1443, 'learning_rate': 2.991522876735154e-06, 'epoch': 3.15}
{'loss': 0.133, 'learning_rate': 2.265259209387867e-06, 'epoch': 3.39}
{'loss': 0.1391, 'learning_rate': 1.6135921418712959e-06, 'epoch': 3.64}
{'loss': 0.1299, 'learning_rate': 1.0542974530180327e-06, 'epoch': 3.88}
{'loss': 0.1414, 'learning_rate': 6.026312439675553e-07, 'epoch': 4.12}
{'loss': 0.1366, 'learning_rate': 2.7091379149682683e-07, 'epoch': 4.36}
{'loss': 0.1349, 'learning_rate': 6.819348298638839e-08, 'epoch': 4.61}
{'loss': 0.1322, 'learning_rate': 0.0, 'epoch': 4.85}
{'train_runtime': 99.7731, 'train_samples_per_second': 6.615, 'train_steps_per_second': 0.2, 'train_loss': 0.2510054260492325, 'epoch': 4.85}
100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 20/20 [01:39<00:00, 4.98s/it]
Trainer.pyfinetune_ds.shfinetune.py我的问题是，有没有办法打印出
labels如有任何帮助，我们将不胜感激。 :)
您可以子类化
Trainercompute_lossfrom transformers import Trainer
class CustomTrainer(Trainer):
def compute_loss(self, model, inputs, return_outputs=False):
# get the labels
labels = inputs.get("labels")
print(labels) # you can use loging
# call the original compute_loss method
return super().compute_loss(model, inputs, return_outputs=return_outputs)
然后您可以使用以下方法创建
trainer # Start trainner
trainer = CustomTrainer(
model=model, tokenizer=tokenizer, args=training_args, **data_module
)