Skip to content

Conversation

@ilyasher
Copy link
Contributor

Overview:

If k8s_model_cache is specified while generating the a dynamo k8s deployment config, set the HF_HOME environment variable to the path where the model cache is mounted, so that HF actually makes use of the mounted directory for caching HF downloads.

Also add a new option, --generator-set K8sConfig.k8s_hf_home=/path/to/hf/home, to override the HF_HOME variable, which could be useful if the HF_HOME dir isn't at the root of the mounted path.

@copy-pr-bot
Copy link

copy-pr-bot bot commented Jan 27, 2026

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@ilyasher ilyasher added the feat label Jan 27, 2026
@ilyasher ilyasher changed the title Add k8s_hf_home option feat: Add k8s_hf_home option Jan 27, 2026
Signed-off-by: Ilya Sherstyuk <isherstyuk@nvidia.com>
@ilyasher ilyasher force-pushed the dev-isherstyuk-add-k8s-hf-home branch from 9c399ec to 7ed0965 Compare January 27, 2026 23:32
@ilyasher ilyasher self-assigned this Jan 27, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants