{ "schema_version": 1, "builder_plan": { "scenario": "You already need task-specific training and want to reduce its memory or compute cost.", "change": "Change only the training technique on a small controlled run after checking prerequisites.", "baseline": "Your current training recipe with the same model, data split, and evaluation.", "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": "Held-out quality, peak memory, training time, and full compute cost to the target quality.", "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": "2610.18683", "title": "ProAS: A Progress-Aware Advantage Shaping Framework for Multi-Turn Agentic Reinforcement Learning", "summary": "ProAS replaces the uniform trajectory-level advantage broadcast of critic-free RL methods such as GRPO with fine-grained credit assignment, using a dynamically built Agentic Reference Graph to mask rewards for invalid actions in successful trajectories and redistribute penalties to early pivotal errors in failed ones. It adds no critic network but requires constructing and maintaining the reference graph during training, which adds implementation complexity and a graph-quality dependency on top of an already compute-intensive RL loop. It can fail if the reference graph is poorly constructed or unrepresentative of the task distribution, and although the authors report robustness to distilled or imperfect graphs, gains were demonstrated only on two benchmarks with small models.", "rationale": "The method is a credible improvement to agentic RL training, but adopting it presupposes an RL fine-tuning workload that is at the edge of feasibility on a single 24 GB GPU, so it is relevant to track rather than to deploy now.", "technique": "Progress-aware advantage shaping for critic-free agentic RL", "family": "treino_eficiente", "recommendation": "observar", "published": "2026-10-13", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.18683", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "ProAS replaces the uniform trajectory-level advantage broadcast of critic-free RL methods such as GRPO with fine-grained credit assignment, using a dynamically built Agentic Reference Graph to mask rewards for invalid actions in successful trajectories and redistribute penalties to early pivotal errors in failed ones. It adds no critic network but requires constructing and maintaining the reference graph during training, which adds implementation complexity and a graph-quality dependency on top of an already compute-intensive RL loop. It can fail if the reference graph is poorly constructed or unrepresentative of the task distribution, and although the authors report robustness to distilled or imperfect graphs, gains were demonstrated only on two benchmarks with small models.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.18683", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "ProAS accelerates training convergence by up to 2.44x over group-based RL baselines and raises success rates to 91.9%/94.3% on ALFWorld and 66.7%/70.8% on WebShop for Qwen2.5-1.5B/7B-Instruct respectively.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.18683", "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": "" }