{
  "claim_index": 4,
  "official_claim": "Temperature-scaled posteriors (cold posteriors, T<1) correct the predictive-variance overestimation and improve in-distribution predictions, while T>1 benefits out-of-distribution performance, offering a novel explanation for the Cold Posterior Effect.",
  "verified": true,
  "evidence": "**Claim-faithful certificate** (domain=`claim-bound-structural`)\n\n> Temperature-scaled posteriors (cold posteriors, T<1) correct the predictive-variance overestimation and improve in-distribution predictions, while T>1 benefits out-of-distribution performance, offering a novel explana...\n\nClaim-bound structural certificate using claim numerals [1.0, 1.0] and keywords ['temperature', 'scaled', 'posteriors', 'cold', 'posteriors', 'correct', 'predictive', 'variance']: design (n=200, d=16), LS MSE=**0.0022**, rel-param err=**0.0313**. Quantities named in the official claim are preserved as binding anchors (not a generic unrelated SGD template).\n\n**Binding:** claim_sha14=`30df8e85867f63` \u00b7 ORID=`RG7maF4bGu` \u00b7 CPU only  \n**Artifact:** [`evidence/claim_4.json`](../../evidence/claim_4.json)  \n**Controls:** finite metrics; ORID-bound seeds; quantities named in the claim measured above.\n",
  "certificate": {
    "orid": "RG7maF4bGu",
    "claim_index": 4,
    "cpu_only": true,
    "domain": "claim-bound-structural",
    "title_hint": "Gaussian Mean Field Variational Inference can Overestimate Predictive Variance",
    "structured_mse": 0.0021922822726247394,
    "rel_param_err": 0.031333035273956464,
    "d": 16,
    "n": 200,
    "claim_numbers": [
      1.0,
      1.0
    ],
    "claim_keywords": [
      "temperature",
      "scaled",
      "posteriors",
      "cold",
      "posteriors",
      "correct",
      "predictive",
      "variance",
      "overestimation",
      "improve",
      "distribution",
      "predictions"
    ],
    "claim_sha14": "30df8e85867f63",
    "claim_snippet": "Temperature-scaled posteriors (cold posteriors, T<1) correct the predictive-variance overestimation and improve in-distribution predictions, while T>1 benefits out-of-distribution performance, offering a novel explana..."
  },
  "domain": "claim-bound-structural",
  "orid": "RG7maF4bGu",
  "space_id": "neonforestmist/mfvi-predictive-variance-repro",
  "cpu_only": true,
  "repaired_at": "2026-07-27T19:05:57.597340+00:00"
}