zagreus-0.4B post-trained to 0.4787 on ITALIC (base 0.28, prior leader 0.372) via committee soft answer-KD, under 10 H100h.
Ruggero Marino Lazzaroni
ruggsea
AI & ML interests
NLP in any form
Recent Activity
updated a model 1 day ago
idealab-cs2/reappraisal-reward-model-v2 updated a model 1 day ago
idealab-cs2/reappraisal-4b-grpo-committee