|
|
Yilong LiEmail / GitHub / Google Scholar / LinkedIn |
|
EMBER: Efficient Memory via Budgeted Evidence Retention for Long-Horizon Agents |
|
Yilong Li, Suman Banerjee, Tong Che
EMBER studies Budgeted Pre-Query Retention for long-horizon agents: an agent ingests a stream before future queries are known, keeps only a fixed budget of source evidence, and later answers from that retained memory without returning to the full raw history. The system learns a retention policy that uses write-time answerability probes to identify source spans likely to support future answers. Instead of storing generic summaries or predicted future questions, EMBER stores evidence capsules: source excerpts paired with retrieval keys that remain searchable at read time. On LongMemEval-RR, EMBER-14B reaches 0.3017 F1 at the 8192-token retained source budget, compared with 0.1765 F1 for the strongest non-EMBER budgeted baseline. The result reframes long-term agent memory as learned evidence retention under strict context and memory budgets. |
AbstractEMBER studies Budgeted Pre-Query Retention for long-horizon agents: an agent ingests a stream before future queries are known, keeps only a fixed budget of source evidence, and later answers from that retained memory without returning to the full raw history. The system learns a retention policy that uses write-time answerability probes to identify source spans likely to support future answers. Instead of storing generic summaries or predicted future questions, EMBER stores evidence capsules: source excerpts paired with retrieval keys that remain searchable at read time. On LongMemEval-RR, EMBER-14B reaches 0.3017 F1 at the 8192-token retained source budget, compared with 0.1765 F1 for the strongest non-EMBER budgeted baseline. The result reframes long-term agent memory as learned evidence retention under strict context and memory budgets. |