1/ Excited to release The ultimate guide to multi-harness RL The same model behaves differently in every agent harness. So we built an open way to train any model with RL on any task set, inside the harnesses people actually use, like Claude Code, Codex, and OpenCode, without changing a single line of harness or training code. Trained across four harnesses, LFM2.5-2.6B went from 42% to 54% with 31% fewer tool calls. 🧵
Adithya S K@adithya_s_k16.1K followers7d ago
0:07Plays on X, where the post was published. Open the original post.
Views on X
98.4KLikes
780Reposts
86Replies
53Post text
1/ Excited to release The ultimate guide to multi-harness RL The same model behaves differently in every agent harness. So we built an open way to train any model with RL on any task set, inside the harnesses people actually use, like Claude Code, Codex, and OpenCode, without changing a single line of harness or training code. Trained across four harnesses, LFM2.5-2.6B went from 42% to 54% with 31% fewer tool calls. 🧵
This entry is part of the launch video directory and the AI Coding launch videos. Counts are read from X and refresh every 15 minutes; the video itself stays on X. See the methodology.