记录torch运行的bug

Traceback (most recent call last):
  File "/mnt2/wsj/table/basetest/test_single.py", line 243, in
    QWen2VL()
  File "/mnt2/wsj/table/basetest/test_single.py", line 116, in QWen2VL
    generated_ids = model.generate(**inputs, max_new_tokens=512)
  File "/home/turing1/miniconda3/envs/MiniCPMV_wsj/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
  File "/home/turing1/miniconda3/envs/MiniCPMV_wsj/lib/python3.10/site-packages/transformers/generation/utils.py", line 2048, in generate
    result = self._sample(
  File "/home/turing1/miniconda3/envs/MiniCPMV_wsj/lib/python3.10/site-packages/transformers/generation/utils.py", line 3044, in _sample
    next_tokens = torch.multinomial(probs, num_samples=1).squeeze(1)
RuntimeError: CUDA error: device-side assert triggered
Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

 试试看只使用一个gpu

    model = Qwen2VLForConditionalGeneration.from_pretrained(
        weight_path,
        torch_dtype=torch.bfloat16,
        device_map="sequential",
    )
    processor = AutoProcessor.from_pretrained(weight_path)

    device = torch.device('cuda:0')    
    text = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
    image_inputs, video_inputs = process_vision_info(messages)
    inputs = processor(
        text=[text],
        images=image_inputs,
        videos=video_inputs,
        padding=True,
        return_tensors="pt",
    ).to(device)
    generated_ids = model.generate(**inputs, max_new_tokens=512)
    generated_ids_trimmed = [
        out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
    ]
    output_text = processor.batch_decode(
        generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
    )