model-sft-dare / README.md
Pritish92's picture
Upload model-sft-dare from Assignment 2
697dd68 verified
metadata
language: en
license: apache-2.0
tags:
  - safety-alignment
  - lora
  - qwen2.5
  - assignment2
base_model: Qwen/Qwen2.5-1.5B-Instruct

Pritish92/model-sft-dare

DARE-sparsified variant of the medical SFT model. Best drop rate selected on validation. Assignment 2 Part 1.

Details

  • Student: 22MF3IM15
  • Base model: Qwen/Qwen2.5-1.5B-Instruct
  • Course: Safety Alignment in LLMs (Assignment 2)