THUDM

/

chatglm2-6b

like 2.03k

Follow

Knowledge Engineering Group (KEG) & Data Mining at Tsinghua University 1,238

When running chatglm2 with bfloat16, it failed.

• Code:

model = AutoModelForCausalLM.from_pretrained("THUDM/chatglm2-6b",torch_dtype=amp_dtype, low_cpu_mem_usage=True,trust_remote_code=True)
tokenizer=AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True)
with torch.no_grad(), torch.autocast(dtype=torch.bfloat16):
    input_ids = tokenizer(prompt, return_tensors="pt").input_ids
        output = model.generate(input_ids, **generate_kwargs)

• Error:
  RuntimeError: Expected query, key, and value to have the same dtype, but got query.dtype: float key.dtype: float and value.dtype: c10::BFloat16 instead.

BTW, it can be run with transformers 4.31.0.