ValueError: Please use the `disk_offload` function instead

Beginners
1

Here is the code:

from transformers import AutoModelForCausalLM, AutoTokenizer
device = "cuda" # the device to load the model onto

model = AutoModelForCausalLM.from_pretrained(
    "Qwen/Qwen2-7B-Instruct",
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2-7B-Instruct")

prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(device)

generated_ids = model.generate(
    model_inputs.input_ids,
    max_new_tokens=512
)
generated_ids = [
    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]

response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]

here is the error:

Traceback (most recent call last):
  File "C:\Users\SunDay\Desktop\NelzGPT\main.py", line 4, in <module>
    model = AutoModelForCausalLM.from_pretrained(
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\SunDay\miniconda3\Lib\site-packages\transformers\models\auto\auto_factory.py", line 564, in from_pretrained
    return model_class.from_pretrained(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\SunDay\miniconda3\Lib\site-packages\transformers\modeling_utils.py", line 4015, in from_pretrained
    dispatch_model(model, **device_map_kwargs)
  File "C:\Users\SunDay\miniconda3\Lib\site-packages\accelerate\big_modeling.py", line 496, in dispatch_model
    raise ValueError(
ValueError: You are trying to offload the whole model to the disk. Please use the `disk_offload` function instead.

what to do?

2

hi @SunDay-s
I don’t know whether there is something wrong with device_map or not.

Can you please try with another small model from Templates for Chat Models ?

And you might find this one helpful Resolving: ValueError: You are trying to offload the whole model to the disk. Please use the `disk_offload` function instead. | by Sridevi Panneerselvam | Medium