sesame

/

csm-1b

like 2.18k

Follow

Sesame 1.42k

Hi everyone,

We’re trying to deploy the sesame/csm-1b model, but running into multiple issues:

1. Hugging Face Inference Deployment:

• The UI displays:
  "This model is not part of our Model Catalog, and does not come with verified configuration. It might behave unexpectedly."
• Additionally, we get a warning:
  "Warning: deploying this model will probably fail because no handler.py file was found in the repository. Try selecting a different model or creating a custom handler."
• As expected, the deployment does not proceed correctly.

2. SageMaker Deployment:

• We tried deploying using the script provided in the Sesame repo.
• While the deployment starts, the endpoint never responds.
• Given the large size of the model, the nodes often time out or need to be manually destroyed.

Has anyone successfully deployed this model either via Hugging Face Inference Endpoints or on SageMaker?

Specific questions:
- Is there a sample handler.py or recommended configuration for this model?
- Are there any known hardware requirements (e.g. GPU spec, RAM) for this model to run stably?
- Has Hugging Face considered adding this model to the verified model catalog, or is that up to the authors?

Any guidance would be much appreciated!

Vishal