Generation with HybridCache fails (affecting Gemma-2) about transformers HOT 2 CLOSED

  File "/home/ss/train_frame/LLaMA-Factory/src/train.py", line 30, in <module>
    main()
  File "/home/ss/train_frame/LLaMA-Factory/src/train.py", line 21, in main
    run_exp()
  File "/home/ss/train_frame/LLaMA-Factory/src/llamafactory/train/tuner.py", line 93, in run_exp
    run_exe(model_args, data_args, training_args, finetuning_args, generating_args, callbacks)
  File "/home/ss/train_frame/LLaMA-Factory/src/llamafactory/train/tuner.py", line 47, in run_exe
    run_sft(model_args, data_args, training_args, finetuning_args, generating_args, callbacks)
  File "/home/ss/train_frame/LLaMA-Factory/src/llamafactory/train/sft/workflow.py", line 107, in run_sft
    predict_results = trainer.predict(dataset, metric_key_prefix="predict", **gen_kwargs)
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/transformers/trainer_seq2seq.py", line 244, in predict
    return super().predict(test_dataset, ignore_keys=ignore_keys, metric_key_prefix=metric_key_prefix)
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/transformers/trainer.py", line 3717, in predict
    output = eval_loop(
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/transformers/trainer.py", line 3826, in evaluation_loop
    losses, logits, labels = self.prediction_step(model, inputs, prediction_loss_only, ignore_keys=ignore_keys)
  File "/home/ss/train_frame/LLaMA-Factory/src/llamafactory/train/sft/trainer.py", line 99, in prediction_step
    loss, generated_tokens, _ = super().prediction_step(  # ignore the returned labels (may be truncated)
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/transformers/trainer_seq2seq.py", line 310, in prediction_step
    generated_tokens = self.model.generate(**generation_inputs, **gen_kwargs)
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/transformers/generation/utils.py", line 1914, in generate
    result = self._sample(
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/transformers/generation/utils.py", line 2651, in _sample
    outputs = self(
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
    return forward_call(*args, **kwargs)
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/accelerate/utils/operations.py", line 822, in forward
    return model_forward(*args, **kwargs)
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/accelerate/utils/operations.py", line 810, in __call__
    return convert_to_fp32(self.model_forward(*args, **kwargs))
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/accelerate/utils/operations.py", line 789, in convert_to_fp32
    return recursively_apply(_convert_to_fp32, tensor, test_type=_is_fp16_bf16_tensor)
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/accelerate/utils/operations.py", line 118, in recursively_apply
    {
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/accelerate/utils/operations.py", line 119, in <dictcomp>
    k: recursively_apply(
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/accelerate/utils/operations.py", line 126, in recursively_apply
    return func(data, *args, **kwargs)
  File "/home/ss/anaconda3-new/envs/train/lib/python3.10/site-packages/accelerate/utils/operations.py", line 781, in _convert_to_fp32
    return tensor.float()
AttributeError: 'HybridCache' object has no attribute 'float'