Failure to use https://huggingface.co/anas-awadalla/mpt-7b model downloading issue? #294
Description
100%|███████████████████████████████████████| 933M/933M [01:59<00:00, 7.79MiB/s]
tokenizer_config.json: 100%
237/237 [00:00<00:00, 53.8kB/s]
tokenizer.json: 100%
2.11M/2.11M [00:00<00:00, 4.15MB/s]
special_tokens_map.json: 100%
99.0/99.0 [00:00<00:00, 24.1kB/s]
Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
config.json: 100%
1.26k/1.26k [00:00<00:00, 321kB/s]
configuration_mpt.py: 100%
9.20k/9.20k [00:00<00:00, 2.21MB/s]
A new version of the following files was downloaded from https://huggingface.co/anas-awadalla/mpt-7b:
- configuration_mpt.py
. Make sure to double-check they do not contain any added malicious code. To avoid downloading new versions of the code file, you can pin a revision.
modeling_mpt.py: 100%
18.4k/18.4k [00:00<00:00, 3.96MB/s]
A new version of the following files was downloaded from https://huggingface.co/anas-awadalla/mpt-7b: - modeling_mpt.py
. Make sure to double-check they do not contain any added malicious code. To avoid downloading new versions of the code file, you can pin a revision.
ImportError Traceback (most recent call last)
Cell In[3], line 2
1 # Foundational Models with 9B parameter with Multimodal Pretraining (MPT) on 7 Billion Tokens
----> 2 model, image_processor, tokenizer = create_model_and_transforms(
3 clip_vision_encoder_path="ViT-L-14",
4 clip_vision_encoder_pretrained="openai",
5 lang_encoder_path="anas-awadalla/mpt-7b",
6 tokenizer_path="anas-awadalla/mpt-7b",
7 cross_attn_every_n_layers=4,
8 cache_dir="/home/mraway/Desktop/src/open_flamingo/open_flamingo" # Defaults to ~/.cache
9 )
11 checkpoint_path = hf_hub_download("openflamingo/OpenFlamingo-9B-vitl-mpt7b", "checkpoint.pt")
12 model.load_state_dict(torch.load(checkpoint_path), strict=False)
File ~/Desktop/src/open_flamingo/open_flamingo/src/factory.py:65, in create_model_and_transforms(clip_vision_encoder_path, clip_vision_encoder_pretrained, lang_encoder_path, tokenizer_path, cross_attn_every_n_layers, use_local_files, decoder_layers_attr_name, freeze_lm_embeddings, cache_dir, **flamingo_kwargs)
60 if text_tokenizer.pad_token is None:
61 # Issue: GPT models don't have a pad token, which we use to
62 # modify labels for the loss.
63 text_tokenizer.add_special_tokens({"pad_token": ""})
---> 65 lang_encoder = AutoModelForCausalLM.from_pretrained(
66 lang_encoder_path,
67 local_files_only=use_local_files,
68 trust_remote_code=True,
69 cache_dir=cache_dir,
70 )
72 # hacks for MPT-1B, which doesn't have a get_input_embeddings method
73 if "mpt-1b-redpajama-200b" in lang_encoder_path:
File ~/anaconda3/envs/flamingo/lib/python3.12/site-packages/transformers/models/auto/auto_factory.py:550, in _BaseAutoModelClass.from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs)
548 if has_remote_code and trust_remote_code:
549 class_ref = config.auto_map[cls.name]
--> 550 model_class = get_class_from_dynamic_module(
551 class_ref, pretrained_model_name_or_path, code_revision=code_revision, **hub_kwargs, **kwargs
552 )
553 _ = hub_kwargs.pop("code_revision", None)
554 if os.path.isdir(pretrained_model_name_or_path):
File ~/anaconda3/envs/flamingo/lib/python3.12/site-packages/transformers/dynamic_module_utils.py:501, in get_class_from_dynamic_module(class_reference, pretrained_model_name_or_path, cache_dir, force_download, resume_download, proxies, token, revision, local_files_only, repo_type, code_revision, **kwargs)
488 # And lastly we get the class inside our newly created module
489 final_module = get_cached_module_file(
490 repo_id,
491 module_file + ".py",
(...)
499 repo_type=repo_type,
500 )
--> 501 return get_class_in_module(class_name, final_module)
File ~/anaconda3/envs/flamingo/lib/python3.12/site-packages/transformers/dynamic_module_utils.py:201, in get_class_in_module(class_name, module_path)
199 name = os.path.normpath(module_path).replace(".py", "").replace(os.path.sep, ".")
200 module_path = str(Path(HF_MODULES_CACHE) / module_path)
--> 201 module = importlib.machinery.SourceFileLoader(name, module_path).load_module()
202 return getattr(module, class_name)
File :649, in _check_name_wrapper(self, name, *args, **kwargs)
File :1176, in load_module(self, fullname)
File :1000, in load_module(self, fullname)
File :537, in _load_module_shim(self, fullname)
File :966, in _load(spec)
File :935, in _load_unlocked(spec)
File :995, in exec_module(self, module)
File :488, in _call_with_frames_removed(f, *args, **kwds)
File ~/.cache/huggingface/modules/transformers_modules/anas-awadalla/mpt-7b/b772e556c8e8a17d087db6935e7cd019e5eefb0f/modeling_mpt.py:18
16 from .configuration_mpt import MPTConfig
17 from .adapt_tokenizer import AutoTokenizerForMOD, adapt_tokenizer_for_denoising
---> 18 from .hf_prefixlm_converter import add_bidirectional_mask_if_missing, convert_hf_causal_lm_to_prefix_lm
19 from .meta_init_context import init_empty_weights
20 from .param_init_fns import MODEL_INIT_REGISTRY, generic_param_init_fn_
File ~/.cache/huggingface/modules/transformers_modules/anas-awadalla/mpt-7b/b772e556c8e8a17d087db6935e7cd019e5eefb0f/hf_prefixlm_converter.py:15
13 import torch
14 from transformers.models.bloom.modeling_bloom import BaseModelOutputWithPastAndCrossAttentions, BloomForCausalLM, BloomModel, CausalLMOutputWithCrossAttentions, CrossEntropyLoss
---> 15 from transformers.models.bloom.modeling_bloom import _expand_mask as _expand_mask_bloom
16 from transformers.models.bloom.modeling_bloom import _make_causal_mask as _make_causal_mask_bloom
17 from transformers.models.bloom.modeling_bloom import logging
ImportError: cannot import name '_expand_mask' from 'transformers.models.bloom.modeling_bloom' (/home/mraway/anaconda3/envs/flamingo/lib/python3.12/site-packages/transformers/models/bloom/modeling_bloom.py)