HFEPX · Eval paper review
Gowthamkumar Nandakishore
Published
Oct 5, 2026
Citations
0
Trust level
High
Usefulness score
77/100 (High)
Extraction confidence
90% (High)
Derived from extracted protocol signals and abstract evidence.
Rater population
Not reported
Signals refreshed
Oct 5, 2026
This paper has strong direct human-feedback and evaluation protocol signal and is suitable as a primary eval pipeline reference.
Use this as a practical starting point for protocol research, then validate against the original paper.
Best use
Primary benchmark and eval reference
Use if you need
A concrete protocol example with enough signal to inform rater workflow design.
What to verify
Validate the exact study setup in the full paper before operational use.
Main weakness
No major weakness surfaced.
Use this as a primary source when designing or comparing eval protocols.
If you are doing eval pipeline work, start here
An open contrastive decision model is near chance as a judge on the hard public benchmarks: Contrastive-LM/CLM-v0.1-8B scores between 0.351 (best- of-four, chance 0.250) and 0.593 (pairwise, chance 0.500), is statistically indistinguishable from coin flipping on RM-Bench and JudgeBench, and answers every HaluEval item with one constant label, matching the trivial always-first baseline at 0.581. Judges with the same parameter count score far higher everywhere: a reward model reaches 0.764 to 0.976 and a generative judge 0.611 to 0.778, and every gap to CLM is significant after Benjamini-Hochberg correction. Two properties do work. Raw confidences are overconfident by up to +0.401, yet one pooled temperature fit on held-out calibration items repairs expected calibration error to at most 0.062, and the repaired confidence ranks the model's own errors above chance on three of six benchmarks. The decision order-flip rate is 0.0002 against 0.2188 for the generative judge, and the length-preference shift is -0.023 against -0.217. The confidence-gated cascade, however, escalates between 0.923 and 1.000 of items to the strong judge at the preregistered 0.97 retention bar: calibrated confidence about a near-chance judge has almost nothing to keep. The design: five public preference benchmarks and one hallucination benchmark with real labels, scored under a preregistration frozen before any test item was seen, against generative, reward-model, and trivial baselines, with per-item predictions released.
These are the protocol signals we could actually recover from the available paper metadata. Use them to decide whether this paper is worth deeper reading.
Pairwise Preference
Directly usable for protocol triage.
"An open contrastive decision model is near chance as a judge on the hard public benchmarks: Contrastive-LM/CLM-v0.1-8B scores between 0.351 (best- of-four, chance 0.250) and 0.593 (pairwise, chance 0.500), is statistically indistinguishable from coin flipping on RM-Bench and JudgeBench, and answers every HaluEval item with one constant label, matching the trivial always-first baseline at 0.581."
Llm As Judge
Includes extracted eval setup.
"An open contrastive decision model is near chance as a judge on the hard public benchmarks: Contrastive-LM/CLM-v0.1-8B scores between 0.351 (best- of-four, chance 0.250) and 0.593 (pairwise, chance 0.500), is statistically indistinguishable from coin flipping on RM-Bench and JudgeBench, and answers every HaluEval item with one constant label, matching the trivial always-first baseline at 0.581."
Calibration
Calibration/adjudication style controls detected.
"Raw confidences are overconfident by up to +0.401, yet one pooled temperature fit on held-out calibration items repairs expected calibration error to at most 0.062, and the repaired confidence ranks the model's own errors above chance on three of six benchmarks."
Rm Bench, Judgebench, Halueval
Useful for quick benchmark comparison.
"An open contrastive decision model is near chance as a judge on the hard public benchmarks: Contrastive-LM/CLM-v0.1-8B scores between 0.351 (best- of-four, chance 0.250) and 0.593 (pairwise, chance 0.500), is statistically indistinguishable from coin flipping on RM-Bench and JudgeBench, and answers every HaluEval item with one constant label, matching the trivial always-first baseline at 0.581."
Calibration error
Useful for evaluation criteria comparison.
"Raw confidences are overconfident by up to +0.401, yet one pooled temperature fit on held-out calibration items repairs expected calibration error to at most 0.062, and the repaired confidence ranks the model's own errors above chance on three of six benchmarks."
An open contrastive decision model is near chance as a judge on the hard public benchmarks: Contrastive-LM/CLM-v0.1-8B scores between 0.351 (best- of-four, chance 0.250) and 0.593 (pairwise, chance 0.500), is statistically indistinguishable from coin flipping on RM-Bench and JudgeBench, and answers every HaluEval item with one constant label, matching the trivial always-first baseline at 0.581.
Based on abstract + metadata only. Check the source paper before making high-confidence protocol decisions.
Human feedback protocol is explicit
Detected: Pairwise Preference
Evaluation mode is explicit
Detected: Llm As Judge
Quality control reporting appears
Detected: Calibration
Benchmark or dataset anchors are present
Detected: Rm-Bench, Judgebench, Halueval
Metric reporting is present
Detected: calibration error
