{ "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": "2609.16091", "title": "Distilling Foundation Models for Agentic What-If Reasoning:Cost, Latency, and Governance in a Hybrid LLM+SLM Architecture", "summary": "The method replaces a TabPFN in-context-learning teacher on the hot inference path with a small distilled feed-forward student for tabular classification in interactive agentic loops. The cost is a per-task distillation and training step, a small quality gap (down to 95.4% accuracy retention on credit-g), and dependence on the teacher's soft targets, which contribute 2.1-7.0 AUC points over hard labels. It can fail when the production data distribution drifts from the distillation benchmarks, since a 17k-parameter student has no capacity to adapt in context and must be retrained.", "rationale": "Distilling a tabular foundation model into a tiny MLP is fully feasible on a single 24 GB GPU or even CPU, directly addresses inference latency in production decision loops, and shows only a modest, measurable accuracy loss.", "technique": "Knowledge distillation of tabular foundation model into compact MLP", "family": "destilacao", "recommendation": "adotar", "published": "2026-09-14", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2609.16091", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": 0, "claims": [ { "claim_id": "claim-01", "statement": "The method replaces a TabPFN in-context-learning teacher on the hot inference path with a small distilled feed-forward student for tabular classification in interactive agentic loops. The cost is a per-task distillation and training step, a small quality gap (down to 95.4% accuracy retention on credit-g), and dependence on the teacher's soft targets, which contribute 2.1-7.0 AUC points over hard labels. It can fail when the production data distribution drifts from the distillation benchmarks, since a 17k-parameter student has no capacity to adapt in context and must be retrained.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.16091", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "The deployed two-head loan pipeline compresses 111.4M parameters to 17,059 (6,532x), retaining 95.4-100.5% accuracy and 96.8-100.0% AUC; the classification head alone compresses 53.2M to 8,546 parameters (6,220x).", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.16091", "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": [ { "anchor": "S3.EGx1-3", "label": "(5)", "section": "3 Background: Distillation versus Fine-Tuning", "role": "loss", "latex": "\\mathcal{L}_{\\text{cls}}(z_{S},z_{T},y) =\\alpha\\,\\mathcal{L}_{\\text{soft}}+(1-\\alpha)\\,\\mathcal{L}_{\\text{hard}}", "mathml": "ℒcls​(zS,zT,y)=α​ℒsoft+(1−α)​ℒhard", "context": "For classification we use the canonical Hinton et al. [12] soft-target objective. Let zS,zT∈ℝC be student and teacher logits, T>0 a softening temperature, y the hard label, and α∈[0,1] the soft/hard mixing weight:", "headline": "", "explanation": "", "objective": "", "symbols": {}, "evidence": "", "evidence_url": "", "source_url": "" }, { "anchor": "S3.E2", "label": "(2)", "section": "3 Background: Distillation versus Fine-Tuning", "role": "proposed_method", "latex": "\\phi^{\\star}=\\arg\\min_{\\phi}\\;\\mathbb{E}_{x\\sim\\mathcal{D}}\\Big[\\mathcal{L}_{\\text{KD}}\\big(g_{\\phi}(x),f_{\\theta_{T}}(x),y\\big)\\Big].", "mathml": "ϕ⋆=arg⁡minϕ​𝔼x∼𝒟​[ℒKD​(gϕ​(x),fθT​(x),y)].", "context": "Distillation trains a separate, compact student architecture gϕ to reproduce the behavior of a fixed teacher fθT with frozen weights, using the teacher’s output distribution as an additional supervisory signal alongside ground-truth labels:", "headline": "", "explanation": "", "objective": "", "symbols": {}, "evidence": "", "evidence_url": "", "source_url": "" }, { "anchor": "S3.E6", "label": "(6)", "section": "3 Background: Distillation versus Fine-Tuning", "role": "loss", "latex": "\\mathcal{L}_{\\text{reg}}(\\hat{y}_{S},\\hat{y}_{T},y)=\\alpha\\cdot\\operatorname{MSE}(\\hat{y}_{S},\\hat{y}_{T})+(1-\\alpha)\\cdot\\operatorname{MSE}(\\hat{y}_{S},y),\\qquad\\alpha=0.7.", "mathml": "ℒreg​(y^S,y^T,y)=α⋅MSE⁡(y^S,y^T)+(1−α)⋅MSE⁡(y^S,y),α=0.7.", "context": "For the continuous regression head there is no softmax to soften, so the analogous soft/hard split uses mean-squared error against the teacher’s point prediction y^T=fθT​(x) and against the true label y:", "headline": "", "explanation": "", "objective": "", "symbols": {}, "evidence": "", "evidence_url": "", "source_url": "" } ], "equations_status": "selected", "core_kind": "formula", "equations_fetched_at": "2026-10-10", "equations_source_url": "", "equations_source_sha256": "", "exposures_reviewed_at": "", "exposures_source_sha256": "" }