LLaMAFactory0.9.1评估猜测报TypeError: GenerationMixin._extract_past_fr
环境：LLaMaFactory:0.9.1
transformers：4.46.1
大模型：GLM-4-9B-chat
错误：
通过LLaMAFactory进行Evaluate& Predict时，报错。
: Traceback (most recent call last):
: File "/data/dms/LLaMA-Factory-0.9.1/src/llamafactory/launcher.py", line 23, in <module>
: launch()
: File "/data/dms/LLaMA-Factory-0.9.1/src/llamafactory/launcher.py", line 19, in launch
: run_exp()
: File "/data/dms/LLaMA-Factory-0.9.1/src/llamafactory/train/tuner.py", line 50, in run_exp
: run_sft(model_args, data_args, training_args, finetuning_args, generating_args, callbacks)
: File "/data/dms/LLaMA-Factory-0.9.1/src/llamafactory/train/sft/workflow.py", line 127, in run_sft
: predict_results = trainer.predict(dataset_module["eval_dataset"], metric_key_prefix="predict", **gen_kwargs)
: File "/data/dms/minconda3/envs/factory091/lib/python3.10/site-packages/transformers/trainer_seq2seq.py", line 259, in predict
: return super().predict(test_dataset, ignore_keys=ignore_keys, metric_key_prefix=metric_key_prefix)
: File "/data/dms/minconda3/envs/factory091/lib/python3.10/site-packages/transformers/trainer.py", line 4042, in predict
: output = eval_loop(
: File "/data/dms/minconda3/envs/factory091/lib/python3.10/site-packages/transformers/trainer.py", line 4158, in evaluation_loop
: losses, logits, labels = self.prediction_step(model, inputs, prediction_loss_only, ignore_keys=ignore_keys)
: File "/data/dms/LLaMA-Factory-0.9.1/src/llamafactory/train/sft/trainer.py", line 121, in prediction_step
: loss, generated_tokens, _ = super().prediction_step( # ignore the returned labels (may be truncated)
: File "/data/dms/minconda3/envs/factory091/lib/python3.10/site-packages/transformers/trainer_seq2seq.py", line 331, in prediction_step
: generated_tokens = self.model.generate(**generation_inputs, **gen_kwargs)
: File "/data/dms/minconda3/envs/factory091/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
: return func(*args, **kwargs)
: File "/data/dms/minconda3/envs/factory091/lib/python3.10/site-packages/transformers/generation/utils.py", line 2215, in generate
: result = self._sample(
: File "/data/dms/minconda3/envs/factory091/lib/python3.10/site-packages/transformers/generation/utils.py", line 3209, in _sample
: model_kwargs = self._update_model_kwargs_for_generation(
: File "/root/.cache/huggingface/modules/transformers_modules/glm-4-9b-chat/modeling_chatglm.py", line 939, in _update_model_kwargs_for_generation
: model_kwargs["past_key_values"] = self._extract_past_from_model_output(
: TypeError: GenerationMixin._extract_past_from_model_output() got an unexpected keyword argument 'standardize_cache_format'
原因：
大模型的transformers版本太低导致，通过config.json文件发现，transformers是4.40.2版本，升级到4.44.0就可以。从
魔搭社区下载最新的配置文件就可，包罗如下7个文件。
https://i-blog.csdnimg.cn/direct/b2df08f737de4821b7a26c8788f49c7a.png
你会发现modeling_chatglm.py的内容有变化，新版本没有standardize_cache_format参数。
https://i-blog.csdnimg.cn/direct/2f0289c853164330a0d794972fe01f8f.png
4.40.2版本https://i-blog.csdnimg.cn/direct/5070baf36b4642c7b3643bae5f691436.png
4.44.0版本
免责声明：如果侵犯了您的权益，请联系站长，我们会及时删除侵权内容，谢谢合作！更多信息从访问主页：qidao123.com:ToB企服之家，中国第一个企服评测及商务社交产业平台。
页:
[1]