Rishidar's picture
Upload folder using huggingface_hub
513788b verified
|
Raw
History Blame Contribute Delete
655 Bytes
metadata
language: en
tags:
  - autoscientist
  - adaption-labs
  - qlora
  - mathcode
  - dpo
base_model: Qwen/Qwen2.5-0.5B-Instruct
datasets:
  - Rishidar/autoscientist-mathcode-dataset

AutoScientist Competition — Mathcode Model

Qwen2.5-0.5B-Instruct adapted for mathcode via Adaption Labs AutoScientist v5: 4-bit QLoRA SFT (r=32, alpha=64) then DPO (beta=0.1) on chosen/rejected pairs.

  • DPO reward accuracy: 0.8181818181818182
  • DPO reward margin: 8.761996030807495

Dataset: Rishidar/autoscientist-mathcode-dataset. Also mirrored on Kaggle: rishidard/autoscientist-mathcode-qlora.