Transformers AutoModelForCasualLM cannot be imported

Question

I am trying to follow this article to use the AutoModelForCasualLM from transformers to generate text with bloom. But I keep getting an error saying that python cannot AutoModelForCasualLM from transformers. I have tried multiple computers and multiple versions of transformers but I always get the following error. (Traceback from latest version of transformers)

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
Cell In[28], line 1
----> 1 from transformers import AutoTokenizer, AutoModelForCasualLM, BloomConfig
      2 from transformers.models.lboom.modeling_bloom import BloomBlock, build_alibi_tensor

ImportError: cannot import name 'AutoModelForCasualLM' from 'transformers' (/mnt/MLDr/venv/lib/python3.10/site-packages/transformers/__init__.py)

code snippet from where the error occurs (first ~10 lines):

import os
import torch
import torch.nn as nn
from collections import OrderedDict
def get_state_dict(shard_num, prefix=None):
    d = torch.load(os.path.join(model_path, f"pytorch_model_{shard_num:05d}-of-00072.bin"))
    return d if prefix is None else OrderedDict((k.replace(prefix, ''), v) for k, v in d.items())
from transformers import AutoTokenizer, AutoModelForCasualLM, BloomConfig
from transformers.models.lboom.modeling_bloom import BloomBlock, build_alibi_tensor
model = "./bloom"
config = BloomConfig.from_pretrained(model_path)
device = 'cpu'

transformers-cli env results:

• transformers version: 4.25.1
• Platform: Linux-5.15.0-58-generic-x86_64-with-glibc2.35
• Python version: 3.10.6
• Huggingface_hub version: 0.11.1
• PyTorch version (GPU?): 1.13.1+cu117 (False)
• Tensorflow version (GPU?): 2.11.0 (False)
• Flax version (CPU?/GPU?/TPU?): not installed (NA)
• Jax version: not installed
• JaxLib version: not installed
• Using GPU in script?: <fill in>
• Using distributed or parallel set-up in script?: <fill in>

Answer 1

I had the same issue, it wasn't related to spelling, what fixed my problem was installing this version of transformers and NOT using the git version:

!pip install transformers==4.20

Answer 2

This is because you are using wrong class name this class name not exist in the version of the Transformers library you are using. The correct class name is AutoModelForCausalLM (note the correct spelling of "Causal"). Try this :

from transformers import AutoTokenizer,AutoModelForCausalLM