Unable to load mistralai/Mistral-7B-Instruct-v0.2. repository not found

[gningue]

Hello,
Did someone have the same problem? I have an error message "repository not found when trying to load the v0.2 model.

python -m vllm.entrypoints.openai.api_server --model mistralai/Mistral-7B-Instruct-v0.2 --max-model-len 32192 --dtype float16 --gpu-memory-utilization 0.6
File "/var/www/vllm/lib/python3.11/site-packages/huggingface_hub/hf_file_system.py", line 789, in _raise_file_not_found raise FileNotFoundError(msg) from err FileNotFoundError: mistralai/Mistral-7B-Instruct-v0.2 (repository not found)

[gningue]

I found a workaround by using a local repository
python -m vllm.entrypoints.openai.api_server --model /user/.cache/huggingface/hub/models--mistralai--Mistral-7B-Instruct-v0.2/snapshots/41b61a33a2483885c981aa79e0df6b32407ed873 --max-model-len 32192 --dtype float16 --gpu-memory-utilization 0.8

[caesar-one]

Mistral (as well as Llama) requires accepting their terms. I fixed doing:
$ pip install huggingface_hub (you can skip this if the package is already installed)
Then:
$ huggingface-cli login
Log in using a token from huggingface.co/settings/tokens

[epignatelli]

The vllm error message should really be improved, then.