Hugging Face

Hugging Face CLI for downloading models, managing datasets, and deploying ML pipelines from the terminal.

35 commands

Browse by Topic

Getting Started

Quick setup and installation

3commands

Auth

Authentication

5commands

Download

Downloading models

8commands

Upload

Uploading to Hub

6commands

Cache

Cache management

6commands

API

Hub API

7commands

Install Hugging Face CLI

$ pip install huggingface_hub

Install the Hugging Face Hub CLI and Python client via pip

Login to Hugging Face

$ huggingface-cli login

Authenticate with Hugging Face using an access token

Install Transformers library

$ pip install transformers

Install the Hugging Face Transformers library for ML models

Login to Hub

$ huggingface-cli login

Authenticate with Hugging Face Hub interactively.

Login with token

$ huggingface-cli login --token $HF_TOKEN

Non-interactive login using access token.

Check login status

$ huggingface-cli whoami

Display current authenticated user.

Logout

$ huggingface-cli logout

Remove stored authentication credentials.

Set token via env

$ export HF_TOKEN=hf_xxxxx

Set access token as environment variable.

Download model

$ huggingface-cli download meta-llama/Llama-2-7b-hf

Download a model to the local cache.

Download to directory

$ huggingface-cli download meta-llama/Llama-2-7b-hf --local-dir ./models

Download model to a specific local directory.

Download specific files

$ huggingface-cli download meta-llama/Llama-2-7b-hf --include "*.safetensors"

Download only files matching a pattern.

Exclude files

$ huggingface-cli download bert-base-uncased --exclude "*.h5" --exclude "*.msgpack"

Skip files matching patterns during download.

Download dataset

$ huggingface-cli download --repo-type dataset imdb

Download a dataset from the Hub.

Download specific revision

$ huggingface-cli download meta-llama/Llama-2-7b-hf --revision v1.0

Download a specific version or branch.

Resume download

$ huggingface-cli download meta-llama/Llama-2-7b-hf --resume-download

Continue an interrupted download.

Download single file

$ huggingface-cli download meta-llama/Llama-2-7b-hf config.json

Download only a specific file from a repo.

Upload folder

$ huggingface-cli upload my-org/my-model ./model-files

Upload a local folder to a Hub repository.

Upload single file

$ huggingface-cli upload my-org/my-model ./model.safetensors

Upload a single file to a repository.

Upload with commit message

$ huggingface-cli upload my-org/my-model ./checkpoint --commit-message "Add v2 checkpoint"

Upload with a descriptive commit message.

Upload dataset

$ huggingface-cli upload my-org/my-dataset ./data --repo-type dataset

Upload files to a dataset repository.

Create private repo

$ huggingface-cli upload my-org/my-model ./files --private --create-pr

Upload to a new private repository via PR.

Upload to subdirectory

$ huggingface-cli upload my-org/my-model ./ckpt --path-in-repo checkpoints/v1

Upload files to a specific path in the repo.

Scan cache

$ huggingface-cli scan-cache

Show all cached models and their sizes.

Delete cache entries

$ huggingface-cli delete-cache

Interactively delete cached models.

Delete specific revisions

$ huggingface-cli delete-cache --delete-revisions abc123 def456

Remove specific cached revisions by hash.

Set cache directory

$ export HF_HOME=/path/to/cache

Configure custom cache location.

Disable cache

$ export HF_HUB_OFFLINE=1

Force offline mode using only cached files.

Show cache path

$ python -c "from huggingface_hub import HfFolder; print(HfFolder.cache_dir)"

Display the current cache directory path.

Get repo info

$ huggingface-cli repo-info meta-llama/Llama-2-7b-hf

Display repository metadata and statistics.

Create repo

$ huggingface-cli repo create my-new-model --type model

Create a new model repository on the Hub.

Create dataset repo

$ huggingface-cli repo create my-dataset --type dataset --private

Create a new private dataset repository.

List files in repo

$ huggingface-cli repo-info meta-llama/Llama-2-7b-hf --files

List all files in a repository.

Search models

$ python -c "from huggingface_hub import HfApi; api=HfApi(); print([m.id for m in api.list_models(filter='llama', limit=5)])"

Search for models by name or filter.

Get model tags

$ python -c "from huggingface_hub import model_info; print(model_info('bert-base-uncased').tags)"

Retrieve model tags and metadata.

Check disk usage

$ huggingface-cli repo-info my-org/my-model --disk-usage

Show repository storage usage.

Discussion

Loading comments...

Hugging Face

Hugging Face CLI for downloading models, managing datasets, and deploying ML pipelines from the terminal.

35 commands

Browse by Topic

Getting Started

Quick setup and installation

3commands

Auth

Authentication

5commands

Download

Downloading models

8commands

Upload

Uploading to Hub

6commands

Cache

Cache management

6commands

API

Hub API

7commands

Install Hugging Face CLI

$ pip install huggingface_hub

Install the Hugging Face Hub CLI and Python client via pip

Login to Hugging Face

$ huggingface-cli login

Authenticate with Hugging Face using an access token

Install Transformers library

$ pip install transformers

Install the Hugging Face Transformers library for ML models

Login to Hub

$ huggingface-cli login

Authenticate with Hugging Face Hub interactively.

Login with token

$ huggingface-cli login --token $HF_TOKEN

Non-interactive login using access token.

Check login status

$ huggingface-cli whoami

Display current authenticated user.

Logout

$ huggingface-cli logout

Remove stored authentication credentials.

Set token via env

$ export HF_TOKEN=hf_xxxxx

Set access token as environment variable.

Download model

$ huggingface-cli download meta-llama/Llama-2-7b-hf

Download a model to the local cache.

Download to directory

$ huggingface-cli download meta-llama/Llama-2-7b-hf --local-dir ./models

Download model to a specific local directory.

Download specific files

$ huggingface-cli download meta-llama/Llama-2-7b-hf --include "*.safetensors"

Download only files matching a pattern.

Exclude files

$ huggingface-cli download bert-base-uncased --exclude "*.h5" --exclude "*.msgpack"

Skip files matching patterns during download.

Download dataset

$ huggingface-cli download --repo-type dataset imdb

Download a dataset from the Hub.

Download specific revision

$ huggingface-cli download meta-llama/Llama-2-7b-hf --revision v1.0

Download a specific version or branch.

Resume download

$ huggingface-cli download meta-llama/Llama-2-7b-hf --resume-download

Continue an interrupted download.

Download single file

$ huggingface-cli download meta-llama/Llama-2-7b-hf config.json

Download only a specific file from a repo.

Upload folder

$ huggingface-cli upload my-org/my-model ./model-files

Upload a local folder to a Hub repository.

Upload single file

$ huggingface-cli upload my-org/my-model ./model.safetensors

Upload a single file to a repository.

Upload with commit message

$ huggingface-cli upload my-org/my-model ./checkpoint --commit-message "Add v2 checkpoint"

Upload with a descriptive commit message.

Upload dataset

$ huggingface-cli upload my-org/my-dataset ./data --repo-type dataset

Upload files to a dataset repository.

Create private repo

$ huggingface-cli upload my-org/my-model ./files --private --create-pr

Upload to a new private repository via PR.

Upload to subdirectory

$ huggingface-cli upload my-org/my-model ./ckpt --path-in-repo checkpoints/v1

Upload files to a specific path in the repo.

Scan cache

$ huggingface-cli scan-cache

Show all cached models and their sizes.

Delete cache entries

$ huggingface-cli delete-cache

Interactively delete cached models.

Delete specific revisions

$ huggingface-cli delete-cache --delete-revisions abc123 def456

Remove specific cached revisions by hash.

Set cache directory

$ export HF_HOME=/path/to/cache

Configure custom cache location.

Disable cache

$ export HF_HUB_OFFLINE=1

Force offline mode using only cached files.

Show cache path

$ python -c "from huggingface_hub import HfFolder; print(HfFolder.cache_dir)"

Display the current cache directory path.

Get repo info

$ huggingface-cli repo-info meta-llama/Llama-2-7b-hf

Display repository metadata and statistics.

Create repo

$ huggingface-cli repo create my-new-model --type model

Create a new model repository on the Hub.

Create dataset repo

$ huggingface-cli repo create my-dataset --type dataset --private

Create a new private dataset repository.

List files in repo

$ huggingface-cli repo-info meta-llama/Llama-2-7b-hf --files

List all files in a repository.

Search models

$ python -c "from huggingface_hub import HfApi; api=HfApi(); print([m.id for m in api.list_models(filter='llama', limit=5)])"

Search for models by name or filter.

Get model tags

$ python -c "from huggingface_hub import model_info; print(model_info('bert-base-uncased').tags)"

Retrieve model tags and metadata.

Check disk usage

$ huggingface-cli repo-info my-org/my-model --disk-usage

Show repository storage usage.

Discussion

Loading comments...