Super-adapter merges weekly harness LoRAs for Qwen3.8-27B, lifting SWE-Bench Pro up 20 points
yb2698 · x · 2026-11-20
- An experimental "super-adapter" for Qwen3.8-27B concatenates all the harness-specific LoRA adapters trained that week into one larger LoRA.
- Results: SWE-Bench Pro (hard 51) Hermes 45%→61%, OpenCode 47%→67%, pi 74%→74%; Terminal-Bench 2.1 gains +4 to +6 points per harness; mean across 6 pairs 55.5%→63.5%.
- With thinking on, it makes 19–48% fewer tool calls depending on the harness. Full recipe blog and open weights on Hugging Face.
More from coding & agent
- Build a 'company brain': one shared knowledge base for your whole team's AI — VibeMarketer_ · 2026-11-21
- LangChain ships four updates: LangSmith Agent Platform, trace data controls, Tool Gateway — LangChain · 2026-11-21
- Stanford CS329Z: Full AI Agents Course Free With Slides, Videos, Homework — stanfordnlp · 2026-11-21
- Grok Bot Gets Its Own Email Addresses to Sign Up and Handle Promos for You — elonmusk · 2026-11-21
- CapCut × ChatGPT plugin turns chat into a video generator and editor — SimplyAnnisa · 2026-11-21
- Princeton & Meta's Strategist evolves inference strategies, beating baselines on 24/30 benchmarks at 39% lower cost — 141_1337 · 2026-11-21