How to save the memory overhead at the beginning of the fine tuning？
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 39.5k
- Forks
- 4.8k
- PR merge metrics
- No merged PRs in 30d
Description
I found that the memory overhead was very high at the beginning of the fine tuning, especially when loading the model by transformers.AutoModelForCausalLM.from_pretrained.
Contributor guide
No contributing guide indexed for this repository
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
The issue names transformers.AutoModelForCausalLM.from_pretrained as the point where memory overhead is observed, but it provides no repository file, test, model, configuration, or reproduction steps. Start by reproducing the fine-tuning setup around model loading and measure memory before and after loading. Done would require a documented cause and an agreed change or workaround that reduces the initial overhead.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100