AI 助理要從過去的記憶中找線索時,只挑和眼前情況最像的內容,常常幫不上忙。
- MemCredit 會測試每段記憶對 AI 助理下一步動作有沒有影響,再用測試結果挑出最有用的內容,放進有限的記憶空間。在 ALFWorld 測試中,記憶上限是 500 個 token 時,MemCredit 的成功率為 86.57%,BM25 搜尋則是 65.67%。
- 每次決定平均使用 152.1 個記憶 token,比 BM25 少 74.5%,而且沒有更動 AI 助理的行動方式。這表示記憶是否有用,不能只看它和目前情況像不像。
AI 伴隨問答
針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:
Interesting insights about agent memory.
The bottom line is that you can't just rely on semantic relevance when picking memories.
Most approaches pick the memory most similar to an agent's current state. But does that help it pick the next action?
This work claims it often does not.
Most retrieved memories have zero effect on the decision, and semantic relevance correlates poorly with usefulness.
MemCredit measures each memory with controlled interventions, recording how the agent's action changes with and without it. A small estimator is trained on those credits. Relevance still collects the candidates, and the estimator decides which ones fit in the token budget.
On ALFWorld with a 500-token memory budget, MemCredit reaches 86.57% success, compared with 65.67% for BM25 retrieval. It uses 152.1 memory tokens per decision, 74.5% fewer than BM25, and the agent policy is unchanged.
Chat with Paper: https://t.co/p1oZT0tQwD