microsoft / microsoft/llm-steer-instruct
Request for Reproducing Figure 3: Steering Vectors and Config Details
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 98
- Forks
- 17
- PR merge metrics
- No merged PRs in 30d
Description
Thank you for sharing this excellent work and open-sourcing the code! I’m trying to reproduce Figure 3 (instruction-following accuracy with and without steering) using the provided codebase, but I’m unsure about the exact configuration used.
Could you kindly share:
-
The layer indices and steering weights applied for each model (e.g., Phi-3, Gemma 2B, etc.).
-
Any config files or scripts that were used to generate the results in Figure 3.
Thanks again for your time and support!
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 requests the layer indices, steering weights, and configuration files or scripts used to produce Figure 3 for each model. Start by locating the repository's experiment entry points and comparing them with the Figure 3 workflow; done means documenting the exact settings needed to reproduce the reported results.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- documentation, machine-learning
- Issue type
- Documentation
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100