sillsdev / sillsdev/silnlp

Add support for APOLLO optimizer

Open
#668 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

pipeline 4: train
Dominant language
Python
Stars
39
Forks
7
Avg merge
1d 9h
Merged PRs (30d)
5

Description

The APOLLO optimizer has been added to Huggingface Transformers. APOLLO is a memory-efficient optimizer with similar performance to AdamW. We will need to add a dependency on the apollo-torch package. APOLLO can be enabled by setting the optim to apollo_adamw. We will also need to automatically set optim_target_modules based on the currently selected model. For more information, see the documentation.

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by locating the dependency declarations and the training or optimizer configuration entry point in silnlp. Verify how Hugging Face Transformers exposes optim=apollo_adamw, then define model-specific optim_target_modules; done means APOLLO can be selected with the dependency installed and targets are set automatically.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.