NVIDIA / NVIDIA/Model-Optimizer
Add MBridge distillation for puzzletron
Open
feature request
- Dominant language
- Python
- Stars
- 3.8k
- Forks
- 604
- Avg merge
- 2d 8h
- Merged PRs (30d)
- 142
Description
Detailed description of the requested feature
After the puzzletron prunes the model, we want a repair phase that will run distillation from the teacher (original model, but possibly any other model) to the pruned model
Timeline
Work in progress. Requires "general" support of heterogeneous models in Mcore. The idea is to patch the mbridge-mcore model creation using the block_configs from anymodel.
This also requires some updates (PRs) to MBridge code as well
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Assessment
This issue has not been assessed yet.