google-research / google-research/t5x
how to use a different optimizer?
Open
- Dominant language
- Python
- Stars
- 3k
- Forks
- 338
- PR merge metrics
- No merged PRs in 30d
Description
There is an excellent example about how to use a different optimizer for a model trained from scratch. But, I am getting an error if I want to use a different optimizer for fine-tuning.
Contributor guide
Research direction
Start by reviewing the existing example for choosing an optimizer when training a model from scratch, then compare it with the fine-tuning setup that produces the error. Reproduce the failure and identify the missing details needed to define the expected behavior; done means fine-tuning supports the alternate optimizer or the documented limitation and usage are clear.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100