Hermes agent can now fine-tune models in-chat via Adaption's AutoScientist plugin
sarahookr · x · 2026-11-18
Sara Hooker (ex-VP of AI research at Cohere) shared that Adaption's plugin is now a verified partner on the Hermes agent, letting you train your own model from inside the chat.
- Data: import from HuggingFace, Kaggle or Google Sheets, or generate synthetic datasets from plain-English prompts; adapt, augment, translate and combine.
- Training: pick a base model, get hyperparameter recommendations, launch a fine-tune with AutoScientist (co-optimizes data and model; claimed to more than double win rates), and watch runs live.
- Caveats: tools run on Adaption's hosted server (browser-authed, not local), consume Adaption credits, data is processed on their side, and the agent quotes a cost upfront but won't block execution.
A linked article covers a full Hermes Agent masterclass: what it is, how to start, how to level up.
More from coding & agent
- New Hermes plugin adaption: fine-tune a model just by asking in chat — sarahookr · 2026-11-18
- Sidehand MCP server drives Windows apps in the background at 2s vs 65s rivals — Agitated_Chair_4977 · 2026-11-18
- UK retailer M&S uses GitHub Copilot multi-agent pipeline to scale retail engineering — marlene_zw · 2026-11-18
- Voice agent team cuts P95 time-to-first-voice by 50% via single-datacenter migration — itsOmSarraf_ · 2026-11-18
- Pro tip: raise reasoning effort to stop GPT-6.2 Sol quitting early on long-horizon tasks — pvncher · 2026-11-18
- HF releases 300MB RL-trained LoRA matching Qwen3.8-27B xhigh accuracy with 8.1× fewer tokens — Disastrous-Work-1632 · 2026-11-18