{ "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": "2610.18690", "title": "Just Respect the Semantics: Image-Token Compression via Pre-LLM Semantic Variable-Resolution Tokenization and In-LLM Relevance Pruning", "summary": "Varest replaces uniform full-resolution image tokenization feeding a VLM with pre-LLM variable-resolution merging guided by object and text localizers, plus learned [REL] registers that prune tokens after early decoder layers. It costs a training run per backbone (a single checkpoint covers multiple budgets) and adds localizer models and merging logic to the inference pipeline, so it is not a drop-in method unless pretrained checkpoints are released. It can fail on tasks requiring fine-grained text or small-object detail if localizers miss regions, and the 7-14% quality gap versus full tokens may be unacceptable for OCR-heavy or high-precision workloads.", "rationale": "The compute savings are substantial and validated across backbones, but the method requires training access to the target VLM and localizer supervision, so the reader should validate it on their own multimodal workload before adopting rather than assume transfer of the reported retention numbers.", "technique": "Semantic variable-resolution image token compression", "family": "memoria_e_contexto", "recommendation": "testar", "published": "2026-10-13", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.18690", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "Varest replaces uniform full-resolution image tokenization feeding a VLM with pre-LLM variable-resolution merging guided by object and text localizers, plus learned [REL] registers that prune tokens after early decoder layers. It costs a training run per backbone (a single checkpoint covers multiple budgets) and adds localizer models and merging logic to the inference pipeline, so it is not a drop-in method unless pretrained checkpoints are released. It can fail on tasks requiring fine-grained text or small-object detail if localizers miss regions, and the 7-14% quality gap versus full tokens may be unacceptable for OCR-heavy or high-precision workloads.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.18690", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "Retains 86.10-92.33% of full-token performance using only 5.56% of the original layer-averaged image-token count (roughly 18x fewer tokens), outperforming 15 baselines on mean retention across three backbones and 13 benchmarks.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.18690", "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": "" }