UnexpectedStatusException: Error for Training job huggingface-qlora-jailbreaks-mistralai–2024-08-22-14-23-02-625: Failed. Reason: AlgorithmError: ExecuteUserScriptError:
ExitCode 1
ErrorMessage “raise EnvironmentError(
OSError: mistralai/Mistral-7B-v0.1 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.”
Command “/opt/conda/bin/python3.10 run_qlora.py --bf16 True --dataset_path /opt/ml/input/data/training --gradient_accumulation_steps 2 --gradient_checkpointing True --learning_rate 0.0002 --logging_steps 10 --lr_scheduler_type constant --max_grad_norm 0.3 --merge_adapters True --model_id mistralai/Mistral-7B-v0.1 --num_train_epochs 1 --output_dir /tmp/run --per_device_train_batch_size 6 --save_strategy epoch --tf32 True --use_flash_attn True --warmup_ratio 0.03”, exit code: 1
Some models on HuggingFace are gated, which essentially means that you must be authorized to use the model.
If you have already been granted access to the model, just make sure you use your HuggingFace API key when running the code. Otherwise, go visit the model page and request for permission. Once you have been granted access, you’ll see a notification like the one in the image below.
“Gated model: You have been granted access to this model.”
Issue"zyronblack/inzaine is not a local folder and is not a valid model identifier listed on ‘Models - Hugging Face’ If this is a private repository, make sure to pass a token having permission to this repo with token or log in with huggingface-cli login."
Tokens allow you to set permissions for each token, but there may be a mistake in the settings.
Or you may be passing the wrong token. Using login() makes it easier to fix.
In addition, there are occasional cases where Hugging Face simply has a bug.