Less Context, More Accuracy: A Bi-Temporal Memory Engine for LLM Agents Where a Lean Retrieved Context Beats the Full History
Quick Answer
Engram is an open-source bi-temporal memory engine that improves LLM accuracy by utilizing a lean context approach, achieving 83.6% on LongMemEval_S with only 9.6k tokens compared to 73.2% for full-context at 79k tokens, while maintaining provenance and reducing costs.
Key Points
- Engram appends lossless episodes without on the critical path.
- Achieves 83.6% accuracy on LongMemEval_S with 9.6k tokens.
- Hybrid read path combines facts and retrieved chunks for better recall.
- Evaluates memory benchmarks with a neutral, in-repo harness.
- Addresses measurement-integrity pitfalls in memory systems.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Long-term memory is the missing layer for LLM agents: across sessions they forget, and the common workaround -- replaying the whole history into the prompt -- is expensive, slow, and, as distractors accumulate, less accurate. Most memory systems win on cost or latency but still lose to the full-context baseline on accuracy, and benchmark numbers are reported on inconsistent, non-reproducible harnesses, so one system appears at wildly different scores across sources. We present Engram, an open-source, dual-process memory engine on a bi-temporal data model. A fast write path appends lossless episodes with no LLM on the critical path; an asynchronous path extracts atomic (subject, predicate, object) facts, builds a bi-temporal knowledge graph, and resolves contradictions without an LLM call per fact -- invalidating, never deleting, so every fact keeps provenance and a supersession chain. A hybrid read path fuses dense, lexical, graph, and recency/salience signals, applies a point-in-time ("as-of") filter, and assembles a compact, provenance-tagged context. On the full 500-question LongMemEval_S, graded by the official category-specific judge, Engram's lean configuration -- answering from a ~9.6k-token retrieved slice, never the full history -- scores 83.6% vs. 73.2% for full-context (+10.4 points, McNemar p < 10^-6) at ~8x fewer tokens (9.6k vs. 79k), with 0/500 errored. The gain needs a hybrid read path: facts alone lose recall, while facts plus retrieved chunks recover detail. We also contribute a neutral, in-repo evaluation harness with the official judge baked in and the full-context baseline in every table, publish the raw per-question logs, and document the measurement-integrity pitfalls (truncation, home-grown judges, full-history leaks) that silently distort memory benchmarks. Every number ships with a command to reproduce it.
| Comments: | 14 pages, 4 figures, 3 tables. Code, reproducible harness, and raw per-question logs: this https URL |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Machine Learning (cs.LG) |
| Cite as: | arXiv:2606.09900 [cs.CL] |
| (or arXiv:2606.09900v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2606.09900 arXiv-issued DOI via DataCite |
Submission history
From: Liuyin Wang [view email]
[v1]
Fri, 5 Jun 2026 11:43:56 UTC (65 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CL
See more →TriAgent: Divergence-Aware Committees for Cost-Efficient Financial Sentiment Analysis
TriAgent introduces a cost-efficient multi-agent system for financial sentiment analysis, combining VADER, FinBERT, and Qwen2.5. It achieves an F1 score of ~0.87 with significant savings of $9.3M/year at a 10M-user scale compared to GPT-4o-mini, while also detecting hallucinations with an AUC of 0.90.