Agentic AI, Harness Engineering, Agentic Memory, Deterministic Agents, LLM Fine-tuning
Developing Agent8088, open-source Agentic harness for making small local model reliable agents:
https://github.com/palindrome-rl/AGENT8088/
Thanks for such a nice and detailed blog on it. Can you also share, how many iterations did this take, and roughly how long end to end? What would be the minimum GPU requirements to train. Also curious how you think the adapter holds up when the harness changes.
Really nice work! Waiting for the next blog before trying it for adapting the model for Agent8088, our open-source harness for making small local models reliable. Also, have you thought about trying model harness Co-Evolution?