Qwen 3.8 27b fine-tune cuts verbose output by up to 40% with little quality loss
julianharris · x · 2026-09-22
- Developer julianharris highlights Qwen-3.8-27b-Swift, a fine-tune of Qwen 3.8 27b that slashes the model's notoriously verbose output by up to 40% without meaningfully affecting quality.
- It ships as a configuration in the awesome-local-ai repo (boxabirds/awesome-local-ai), which offers one-command installers that run models locally behind an OpenAI-compatible API with a coding agent (opencode) already wired up — including Ubuntu 24GB (llama.cpp) and macOS 64GB (MLX) scripts for the Swift fine-tune.
- The repo also contains benchmarks, model combination configs, Qwen 3.8 experiment demos, and other variants like flash-next, plus the author's article on local AI and fine-tunes.
More from coding & agent
- Pigment Agents now reads and posts in Slack, inheriting user permissions — mattturck · 2026-11-12
- Map of the agent economy: 50+ startups across payments, spend control, identity, memory — venturetwins · 2026-11-12
- OpenCode launches Browse, an on-device browser agent with local logins — Saul_Loveman · 2026-11-12
- VentureTwins maps the B2A economy: dozens of agent infra and app companies in first series installment — venturetwins · 2026-11-12
- Google releases 76-page 'Agents Companion' whitepaper on agent engineering — mdancho84 · 2026-11-12
- Spinning up containers for 20-line agent tools is backwards: Safescript compiles provable, 50ms-deploy apps — uriwa · 2026-11-12