RLHF take-home aligning a small open modelCodingA take-home over a few days: align a small open model with preference data and a parameter-efficient adapter on a single GPU, push the result to the Hugging Face Hub, and write up…
Hugging Face take-home interview
Round 3 of 6 · 1 question reported
- Format
- A task drawn from real work: fix a bug in a real repository, add a feature to a library, a dataset contribution, a Spaces demo; ML roles may get a small training or RLHF task
- They grade
- Working code, tests, a clear README and honest notes on tradeoffs and what you cut
- Heads-up
- Sources disagree on size: one says 2 to 4 hours, another says there is no timer at all. Ask how long it should take.