deepspeedai / deepspeedai/DeepSpeed
Add support for autotuning PEFT models
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 43.1k
- Forks
- 5k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 112
Description
I was trying autotuner with prompt tuning and it counted the parameters of the whole model and deducted that it doesn't fit in memory.
Describe the solution you'd like
I want autotuner to consider that there might be some frozen parameters
Describe alternatives you've considered
I just tuned the hyperparameters by hand.
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.
Research direction
Start by locating the autotuner logic that counts model parameters and inspect how it determines memory fit. Reproduce the issue with a prompt-tuning or otherwise frozen-parameter model, then verify that autotuning accounts only for parameters that require tuning and still selects valid hyperparameters.
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
- 35/100