Most language model fine-tuning research solves a clean problem: take some text, predict the next token, collect a scalar loss, backpropagate, update weights. This works beautifully for single-turn tasks. But real AI agents don’t work this way. They make decisions across multiple timesteps. Environments branch into unexpected futures. Feedback arrives o…
Can evolutionary search train long-horizon agents without the usual GPU-heavy RL machinery?
Agentic ESOpt: Fine-Tuning Long-Horizon LLM Agents with Minimal GPU Requirements
