How to use gated models?

[premissa72]

I am testing some language models in my research. I have a problem with gated models specifically with the meta-llama/Llama-2-7b-hf. As I can only use the environment provided by the university where I work, I use docker for this purpose. With the not gated models, I had no problems but trying the llama model I am getting an error even in the case I provide my access token. The CLI cannot be used in this case, but I have managed to log in via the notebook. What is the issue? Please help with the correct process.

from huggingface_hub import login
login(token = 'my token')

It was successfull.

# Load the tokenizer
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-hf", token='my token')

It has failed even I tried use_auth_token=True

The error message:

---------------------------------------------------------------------------
HTTPError                                 Traceback (most recent call last)
File ~/.local/lib/python3.10/site-packages/huggingface_hub/utils/_errors.py:261, in hf_raise_for_status(response, endpoint_name)
    260 try:
--> 261     response.raise_for_status()
    262 except HTTPError as e:

File /opt/conda/lib/python3.10/site-packages/requests/models.py:1021, in Response.raise_for_status(self)
   1020 if http_error_msg:
-> 1021     raise HTTPError(http_error_msg, response=self)

HTTPError: 403 Client Error: Forbidden for url: https://huggingface.co/meta-llama/Llama-2-7b-hf/resolve/main/tokenizer_config.json

The above exception was the direct cause of the following exception:

GatedRepoError                            Traceback (most recent call last)
File ~/.local/lib/python3.10/site-packages/transformers/utils/hub.py:417, in cached_file(path_or_repo_id, filename, cache_dir, force_download, resume_download, proxies, use_auth_token, revision, local_files_only, subfolder, repo_type, user_agent, _raise_exceptions_for_missing_entries, _raise_exceptions_for_connection_errors, _commit_hash)
    415 try:
    416     # Load from URL or cache if already cached
--> 417     resolved_file = hf_hub_download(
    418         path_or_repo_id,
    419         filename,
    420         subfolder=None if len(subfolder) == 0 else subfolder,
    421         repo_type=repo_type,
    422         revision=revision,
    423         cache_dir=cache_dir,
    424         user_agent=user_agent,
    425         force_download=force_download,
    426         proxies=proxies,
    427         resume_download=resume_download,
    428         use_auth_token=use_auth_token,
    429         local_files_only=local_files_only,
    430     )
    432 except RepositoryNotFoundError:

File ~/.local/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py:118, in validate_hf_hub_args.<locals>._inner_fn(*args, **kwargs)
    116     kwargs = smoothly_deprecate_use_auth_token(fn_name=fn.__name__, has_token=has_token, kwargs=kwargs)
--> 118 return fn(*args, **kwargs)

File ~/.local/lib/python3.10/site-packages/huggingface_hub/file_download.py:1195, in hf_hub_download(repo_id, filename, subfolder, repo_type, revision, library_name, library_version, cache_dir, local_dir, local_dir_use_symlinks, user_agent, force_download, force_filename, proxies, etag_timeout, resume_download, token, local_files_only, legacy_cache_layout)
   1194 try:
-> 1195     metadata = get_hf_file_metadata(
   1196         url=url,
   1197         token=token,
   1198         proxies=proxies,
   1199         timeout=etag_timeout,
   1200     )
   1201 except EntryNotFoundError as http_error:
   1202     # Cache the non-existence of the file and raise

File ~/.local/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py:118, in validate_hf_hub_args.<locals>._inner_fn(*args, **kwargs)
    116     kwargs = smoothly_deprecate_use_auth_token(fn_name=fn.__name__, has_token=has_token, kwargs=kwargs)
--> 118 return fn(*args, **kwargs)

File ~/.local/lib/python3.10/site-packages/huggingface_hub/file_download.py:1541, in get_hf_file_metadata(url, token, proxies, timeout)
   1532 r = _request_wrapper(
   1533     method="HEAD",
   1534     url=url,
   (...)
   1539     timeout=timeout,
   1540 )
-> 1541 hf_raise_for_status(r)
   1543 # Return

File ~/.local/lib/python3.10/site-packages/huggingface_hub/utils/_errors.py:277, in hf_raise_for_status(response, endpoint_name)
    274     message = (
    275         f"{response.status_code} Client Error." + "\n\n" + f"Cannot access gated repo for url {response.url}."
    276     )
--> 277     raise GatedRepoError(message, response) from e
    279 elif error_code == "RepoNotFound" or response.status_code == 401:
    280     # 401 is misleading as it is returned for:
    281     #    - private and gated repos if user is not authenticated
    282     #    - missing repos
    283     # => for now, we process them as `RepoNotFound` anyway.
    284     # See https://gist.github.com/Wauplin/46c27ad266b15998ce56a6603796f0b9

GatedRepoError: 403 Client Error. (Request ID: Root=1-64f2ec62-606f496c0d95f5a34a4a0d0f;01831139-7411-49fd-9c24-89b875b4dcf0)

Cannot access gated repo for url https://huggingface.co/meta-llama/Llama-2-7b-hf/resolve/main/tokenizer_config.json.
Your request to access model meta-llama/Llama-2-7b-hf is awaiting a review from the repo authors.

During handling of the above exception, another exception occurred:

OSError                                   Traceback (most recent call last)
Cell In[25], line 2
      1 # Load the tokenizer
----> 2 tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-hf", token='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', use_auth_token=True)

File ~/.local/lib/python3.10/site-packages/transformers/models/auto/tokenization_auto.py:652, in AutoTokenizer.from_pretrained(cls, pretrained_model_name_or_path, *inputs, **kwargs)
    649     return tokenizer_class.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
    651 # Next, let's try to use the tokenizer_config file to get the tokenizer class.
--> 652 tokenizer_config = get_tokenizer_config(pretrained_model_name_or_path, **kwargs)
    653 if "_commit_hash" in tokenizer_config:
    654     kwargs["_commit_hash"] = tokenizer_config["_commit_hash"]

File ~/.local/lib/python3.10/site-packages/transformers/models/auto/tokenization_auto.py:496, in get_tokenizer_config(pretrained_model_name_or_path, cache_dir, force_download, resume_download, proxies, use_auth_token, revision, local_files_only, subfolder, **kwargs)
    434 """
    435 Loads the tokenizer configuration from a pretrained model tokenizer configuration.
    436 
   (...)
    493 tokenizer_config = get_tokenizer_config("tokenizer-test")
    494 ```"""
    495 commit_hash = kwargs.get("_commit_hash", None)
--> 496 resolved_config_file = cached_file(
    497     pretrained_model_name_or_path,
    498     TOKENIZER_CONFIG_FILE,
    499     cache_dir=cache_dir,
    500     force_download=force_download,
    501     resume_download=resume_download,
    502     proxies=proxies,
    503     use_auth_token=use_auth_token,
    504     revision=revision,
    505     local_files_only=local_files_only,
    506     subfolder=subfolder,
    507     _raise_exceptions_for_missing_entries=False,
    508     _raise_exceptions_for_connection_errors=False,
    509     _commit_hash=commit_hash,
    510 )
    511 if resolved_config_file is None:
    512     logger.info("Could not locate the tokenizer configuration file, will try to use the model config instead.")

File ~/.local/lib/python3.10/site-packages/transformers/utils/hub.py:433, in cached_file(path_or_repo_id, filename, cache_dir, force_download, resume_download, proxies, use_auth_token, revision, local_files_only, subfolder, repo_type, user_agent, _raise_exceptions_for_missing_entries, _raise_exceptions_for_connection_errors, _commit_hash)
    417     resolved_file = hf_hub_download(
    418         path_or_repo_id,
    419         filename,
   (...)
    429         local_files_only=local_files_only,
    430     )
    432 except RepositoryNotFoundError:
--> 433     raise EnvironmentError(
    434         f"{path_or_repo_id} is not a local folder and is not a valid model identifier "
    435         "listed on 'https://huggingface.co/models'\nIf this is a private repository, make sure to "
    436         "pass a token having permission to this repo with `use_auth_token` or log in with "
    437         "`huggingface-cli login` and pass `use_auth_token=True`."
    438     )
    439 except RevisionNotFoundError:
    440     raise EnvironmentError(
    441         f"{revision} is not a valid git identifier (branch name, tag name or commit id) that exists "
    442         "for this model name. Check the model page at "
    443         f"'https://huggingface.co/{path_or_repo_id}' for available revisions."
    444     )

OSError: meta-llama/Llama-2-7b-hf is not a local folder and is not a valid model identifier listed on 'https://huggingface.co/models'
If this is a private repository, make sure to pass a token having permission to this repo with `use_auth_token` or log in with `huggingface-cli login` and pass `use_auth_token=True`.

[UserDAN]

I have the same issue

[ReatKay]

@premissa72

tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-hf", token='my token')

I had the same issues when I tried the Llama-2 model with a token passed through code. I gave up after while using cli.
But what I see from your error:
**

“Your request to access model meta-llama/Llama-2-7b-hf is awaiting a review from the repo authors.”

**
I have an assumption. I suspect some auth response caching issues or - less likely - some extreme response caching expiration time values. As you said you could login via notebook (and access the repo, I assume?), the caching seems to be client based. While I had issues with token auth too, now it works for with token (tried from secondary environment).

[UserDAN]

when can I get the approval from hugging face it has been two days

if anyone know to can I contact them please reply to me

[ReatKay]

The approval does not come from hugging face, it will come from the repo owner, in this case meta. I assume during weekends their repo owner doesnt work

[UserDAN]

thank you for the replay
is there any way to contact them??

[ReatKay]

just wait. I assume they’re US based - so depending on their loc their work week just starts or is going to start. And probably over the weekend there is a number of requests gathering up to be checked.

And there’s always possibility that the responsible person isn’t around on day

[UserDAN]

thank you for your replays

while I am waiting I tried to used this free API

but when I run it in python it gave me this error:
{‘error’: ‘Model requires a Pro subscription; check out Hugging Face – Pricing to learn more. Make sure to include your HF token in your query.’}

Do you know why?

[rakaar]

An alternative way is to download LLAMA weights from Meta website and load the model from the downloaded weights

1. Fill the form on Meta’s website - Download Llama
2. You will promptly get a email with a URL
3. Go the repo in the email, and you will find a download.sh file.
4. Run the download.sh in your machine and when asked paste the link that you received in email.
5. Once you have the weights, there is a llama folder, which has params.json. Copy the file as config.json and change the parameter “vocab_size”. Default is -1. It can be changed to 32000.
6. Use this piece of code to load the model

from transformers import AutoModelForCausalLM, AutoConfig

llama_config = AutoConfig.from_pretrained("/home/rka/code/llama-2-7b-chat")
llama_model = AutoModelForCausalLM.from_config(llama_config)

# for cuda
import torch
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
llama_model.to(device)

[anshul2110]

from transformers import AutoModelForCausalLM, AutoConfig
import torch

Load the model configuration from a local directory

llama_config = AutoConfig.from_pretrained(“/home/rka/code/llama-2-7b-chat”)

Load the model from the configuration

llama_model = AutoModelForCausalLM.from_config(llama_config)

Move the model to a CUDA device if available

device = torch.device(“cuda” if torch.cuda.is_available() else “cpu”)
llama_model.to(device)

Load the tokenizer

from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained(“/home/rka/code/llama-2-7b-chat”)

Define a function to generate text using the model

def generate_text(input_text, max_length=100):
inputs = tokenizer(input_text, return_tensors=“pt”)
inputs = {k: v.to(device) for k, v in inputs.items()}
output = llama_model.generate(**inputs, max_length=max_length)
response = tokenizer.decode(output[0], skip_special_tokens=True)
return response

Test the model

input_text = “Who are you?”
response = generate_text(input_text)
print(response)

---

HFValidationError Traceback (most recent call last)
/usr/local/lib/python3.10/dist-packages/transformers/utils/hub.py in cached_file(path_or_repo_id, filename, cache_dir, force_download, resume_download, proxies, token, revision, local_files_only, subfolder, repo_type, user_agent, _raise_exceptions_for_gated_repo, _raise_exceptions_for_missing_entries, _raise_exceptions_for_connection_errors, _commit_hash, **deprecated_kwargs)
401 # Load from URL or cache if already cached
→ 402 resolved_file = hf_hub_download(
403 path_or_repo_id,

7 frames
HFValidationError: Repo id must be in the form ‘repo_name’ or ‘namespace/repo_name’: ‘/home/rka/code/llama-2-7b-chat’. Use repo_type argument if needed.

The above exception was the direct cause of the following exception:

OSError Traceback (most recent call last)
/usr/local/lib/python3.10/dist-packages/transformers/utils/hub.py in cached_file(path_or_repo_id, filename, cache_dir, force_download, resume_download, proxies, token, revision, local_files_only, subfolder, repo_type, user_agent, _raise_exceptions_for_gated_repo, _raise_exceptions_for_missing_entries, _raise_exceptions_for_connection_errors, _commit_hash, **deprecated_kwargs)
464 raise EnvironmentError(f"There was a specific connection error when trying to load {path_or_repo_id}:\n{err}“)
465 except HFValidationError as e:
→ 466 raise EnvironmentError(
467 f"Incorrect path_or_model_id: ‘{path_or_repo_id}’. Please provide either the path to a local folder or the repo_id of a model on the Hub.”
468 ) from e

OSError: Incorrect path_or_model_id: ‘/home/rka/code/llama-2-7b-chat’. Please provide either the path to a local folder or the repo_id of a model on the Hub.

getting this error any help ?