Super-Adapter for Qwen3-27B: Merging All LoRAs Lifts SWE-Bench Pro to 67%
yb2698 · x · 2026-11-20
The author released an experimental "super-adapter" for Qwen3.8-27B: a bigger LoRA that concatenates all harness-specific adapters trained this week into one.
Results:
- SWE-Bench Pro (hard 51): Hermes Agent 45%→61%, OpenCode 47%→67%, pi 74%→74%
- Terminal-Bench 2.1: +4 to +6 points across harnesses
- Beats the thinking model with 19-48% fewer tool calls in thinking mode
Full recipe, experiments, and open questions are shared in a blog post.
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