RESEARCH · RESEARCH · #1635
AMBER: append-only memory bank for long-horizon web agents
The paper (arXiv:2610.07118v1) introduces AMBER, an append-only memory bank that an agent learns to write to jointly with reasoning and acting, with an append-only rule that guarantees retention. Trained end-to-end with reinforcement learning (without extensive supervised fine-tuning), AMBER improves average success on WebArena Lite by 4.09 percentage points, raises the fraction of tasks solved in five repeated runs by 4.8 percentage points, and matches an overwrite-memory baseline that used more expensive curated supervision while remaining token-efficient.
KEY POINTS
- The paper (arXiv:2610.07118v1) introduces AMBER, an append-only memory bank that an agent learns to write to jointly with reasoning and acting, with an append-only rule that guarantees retention.
- Trained end-to-end with reinforcement learning (without extensive supervised fine-tuning), AMBER improves average success on WebArena Lite by 4.09 percentage points, raises the fraction of tasks solved in five repeated runs by 4.8 percentage points, and matches an overwrite-memory baseline that used more expensive curated supervision while remaining token-efficient.
- Append-only memory prevents deletion of critical facts and corrective feedback, making long-horizon interactive agents more reliable without costly supervised data.
WHY IT MATTERS
Append-only memory prevents deletion of critical facts and corrective feedback, making long-horizon interactive agents more reliable without costly supervised data.