{ "schema_version": 1, "builder_plan": { "scenario": "A support or data assistant repeats work across related customer requests.", "change": "Try the paper's memory change in a replay of past tasks, with separate memory per customer.", "baseline": "Your current memory pipeline plus a no-memory control using the same agent.", "steps": [ "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.", "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.", "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.", "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test." ], "measure": "Task success, stale memories, retrieval misses, response latency, and all memory maintenance costs.", "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation." }, "builder_review": null, "arxiv_id": "2610.11287", "title": "REMORY: Learning Residual Memory for Context Compaction", "summary": "REMORY replaces the plain-text summary used in context compaction with a summary plus a bounded sequence of learned soft memory tokens, appended as a residual along the sequence dimension to recover information the summary dropped. The cost is training a separate memory network per backbone LLM against a frozen model's continuation objective, plus the inference-time overhead of generating the tokens, which demands gradient-level access to the target model during training and is impractical for API-only or 27B-class models on a single 24 GB GPU. It can fail when deployed on a backbone or domain different from the training setup, since the soft tokens are model-specific and their benefit is validated mainly on summarization-attribution and two agent benchmarks rather than general workloads.", "rationale": "The method is promising for long-horizon agent context management, but adopting it requires training a backbone-specific memory module against a frozen LLM, which exceeds the reader's constrained infrastructure and API-based workflow for the models demonstrated.", "technique": "Residual soft memory tokens for context compaction", "family": "memoria_e_contexto", "recommendation": "observar", "published": "2026-10-08", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.11287", "independent_implementations": 3, "total_implementations": 4, "stars": 169, "citations": 0, "claims": [ { "claim_id": "claim-01", "statement": "REMORY replaces the plain-text summary used in context compaction with a summary plus a bounded sequence of learned soft memory tokens, appended as a residual along the sequence dimension to recover information the summary dropped. The cost is training a separate memory network per backbone LLM against a frozen model's continuation objective, plus the inference-time overhead of generating the tokens, which demands gradient-level access to the target model during training and is impractical for API-only or 27B-class models on a single 24 GB GPU. It can fail when deployed on a backbone or domain different from the training setup, since the soft tokens are model-specific and their benefit is validated mainly on summarization-attribution and two agent benchmarks rather than general workloads.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.11287", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "On SummHay, approaches the full-context joint score using only 5.2% of input positions, with improved source attribution at nearly unchanged insight coverage; Qwen3.8-27B and GLM-5.3-Flash show fewer repeated tool outputs and tool errors on BrowseComp and Terminal-Bench 2.1.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.11287", "source_page": null, "source_excerpt": "" } ], "exposure_map": [ { "dimension": "quality", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compute", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "latency", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "operations", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compatibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "security", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "data_and_training", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "reproducibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" } ], "risks": [], "minimum_test": [], "open_questions": [], "independent_tests": [], "lab_reproductions": [], "report_available": false, "validation_tier": "unknown", "evidence_tier": "unknown", "infrastructure_basis": "unknown", "equations": [], "equations_status": "not_fetched", "core_kind": "", "equations_fetched_at": "", "equations_source_url": "", "equations_source_sha256": "", "exposures_reviewed_at": "", "exposures_source_sha256": "" }