feature request: support 4d attention masks
Open
Nobody has claimed this yet.
- Dominant language
- C++
- Stars
- 4.7k
- Forks
- 536
- Avg merge
- 12h 12m
- Merged PRs (30d)
- 4
Description
this is a Feature Request to implement custom 4D mask for Llama (and possibly any other model) similar to https://github.com/huggingface/transformers/pull/27539
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 reviewing the referenced Hugging Face Transformers pull request 27539 to understand the intended 4D attention-mask behavior. Then trace how CTranslate2 handles attention masks for Llama and determine the scope for supporting custom masks in other models. Done means custom 4D masks work correctly for Llama, with any broader support clearly defined.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100