{ "schema_version": 1, "builder_plan": { "scenario": "A narrow, repeated product task may not need your largest model.", "change": "Check whether a released student model fits the task before generating training data.", "baseline": "Your current model and an off-the-shelf small model on the same held-out tasks.", "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": "Student quality, serving cost, teacher calls, training cost, and failures on unusual inputs.", "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": "2609.40253", "title": "ComputerSD: Online Self-Distillation from Real-Time Feedback for Computer-Use Agents", "summary": "ComputerSD replaces outcome-only GRPO training of computer-use agents with joint token-level self-distillation (using real-time GUI feedback from a fine-tuned analyzer) plus trajectory-level GRPO. It costs an additional fine-tuned GUI analyzer model, privileged rescoring per step, and a fully asynchronous online training framework that requires an executable GUI environment — infrastructure beyond a single 24 GB GPU for the 8B backbones used. Failure modes include misalignment between the analyzer's guidance and the student's evolving state, analyzer errors propagating as flawed supervision, and modest absolute gains (1.9–4.1 points) that may not generalize beyond the evaluated backbones.", "rationale": "The method requires online RL infrastructure, an executable GUI environment, and training capacity for 8B vision-language models, none of which fit the reader's single 24 GB GPU and limited budget, though the results are worth tracking for future capability.", "technique": "Online self-distillation from GUI feedback for CUAs", "family": "destilacao", "recommendation": "observar", "published": "2026-09-30", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2609.40253", "independent_implementations": 0, "total_implementations": 1, "stars": 2, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "ComputerSD replaces outcome-only GRPO training of computer-use agents with joint token-level self-distillation (using real-time GUI feedback from a fine-tuned analyzer) plus trajectory-level GRPO. It costs an additional fine-tuned GUI analyzer model, privileged rescoring per step, and a fully asynchronous online training framework that requires an executable GUI environment — infrastructure beyond a single 24 GB GPU for the 8B backbones used. Failure modes include misalignment between the analyzer's guidance and the student's evolving state, analyzer errors propagating as flawed supervision, and modest absolute gains (1.9–4.1 points) that may not generalize beyond the evaluated backbones.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.40253", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "On OSWorld-Verified, ComputerSD outperforms outcome-only GRPO by 1.9 points on Qwen3-VL-8B-Thinking and 4.1 points on EvoCUA-8B.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.40253", "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": "" }