Event
Proposes the TokenWorth online memory framework
1 reports · 1 sources
The paper proposes TokenWorth, a lightweight online memory framework that learns query-relevant memory value from sparse execution feedback and uses it uniformly for memory retrieval and bounded memory management. Across three LifelongAgentBench environments and under both local and API deployments, TokenWorth achieves the highest task success rate among the evaluated methods while producing the fewest output tokens; the code has been released publicly on GitHub.
Report timeline
- TokenWorth: Learning which memories are worth keeping under resource constraints
The paper proposes TokenWorth, a lightweight online memory framework that learns query-relevant memory value from sparse execution feedback and applies it consistently to memory retrieval and bounded memory management. Across three LifelongAgentBench environments and in both local and API-based deployments, TokenWorth achieves the highest task success rate among the evaluated methods while using the fewest output tokens. The code is publicly available at https://github.com/AIGNLAI/TokenWorth.
HuggingFace Daily Papers ·