Agentic ESOpt: Fine-Tuning Long-Horizon LLM Agents with Minimal GPU Memory
_akhaliq · x · 2026-08-20
Agentic ESOpt is a new framework designed to reduce the high GPU memory costs of training long-horizon LLM agents. It replaces traditional backpropagation with evolution strategies, enabling full-parameter optimization with just inference-level memory requirements. The method demonstrates strong performance gains on benchmarks like WebArena-Lite.
Related event: Agentic ESOpt Fine-Tunes Long-Horizon LLM Agents with Minimal GPU Memory(2 posts)→
More from coding & agent
- Benchmark: Pi Agent wins tasks, DeepSeek Harness wins cost — mgostIH · 2026-08-20
- Omnigent 0.10.0 released: Multi-sandbox support and Devin integration — matei_zaharia · 2026-08-20
- Cursor AI releases subscriptions and four other agent features — mattyp · 2026-08-20
- Agent Security: Policy-Driven Gateway for Tool Discovery — Strange_Profit_8129 · 2026-08-20
- Struggling with Memory Silos Across AI Tools: Permissions and Deduplication — the-cybersapien · 2026-08-20
- All-Gemini agent crew: 2.53x faster cognitive cycles, zero repair cycles — leebase65 · 2026-08-20