[NeurIPS'2026] Position Bias in Mamba and Hybrid Language Models | Evidence Position Bias Across Sequence Mixers in Long-Context Question Answering
-
Updated
Aug 22, 2026 - Python
[NeurIPS'2026] Position Bias in Mamba and Hybrid Language Models | Evidence Position Bias Across Sequence Mixers in Long-Context Question Answering
Seven decision models, measured as the candidate list grows, the option order changes, and the wrong answers stop being obvious.
ACL 2026 | CapCal: content-agnostic probability calibration for de-biasing listwise rerankers.
Do LLM judges know when they're wrong? Three open-weight judges (Qwen2.5-7B, kev-8b, auto-j-13b) graded against MT-Bench human votes: calibration, position and padding attacks, and what auto-accepting confident verdicts would cost. Interactive site included.
Learning-to-rank lab: listwise models, position bias, calibration, and NDCG evaluation
Option-reversal control for paired binary vision–language benchmarks: estimates a model's answer-slot bias and perceptual accuracy from one extra inference pass per question, attributes each failure to slot or perception, and reports scores against the correct chance floors. Ongoing project; code only.
position bias in LLM judges is worse than people think: ask a small instruct model to choose 1 of 2 items. 81% of the answer will be based on which slot the item was in, not which it was. swap the order and the answer flips 78% of the time. an inconclusive cognitive dissonance experiment
Position-bias-aware ranking: estimating and correcting position bias to optimize Earnings Per Visitor (EPV) · simulation study · IPW & propensity modeling · Python
Blind side-by-side human eval of two model responses on a weighted rubric — randomized panes to kill position bias, bootstrap CIs on the margin, inter-rater reliability. An eval you can't audit is a vote, not a measurement.
SSIT: a label-free, gold-free test for whether LLM-judge position x verbosity bias corrections actually compose. No human labels, no model of the judge. Code, 7-judge/6-family pilot data, and pre-registered protocol for the NeurIPS 2026 JUDGe workshop paper.
Measure position/verbosity/assertiveness bias in an LLM-as-judge (Claude) by judging pairs in both orders. Finding: no position bias, but 75% verbosity bias and 100% assertiveness bias — judge scores gameable by length + tone.
Two kinds of saturation: why LLM-judge order bias is hard to measure — essay, Lean proofs (0 sorry), and a reproducible dispersion measurement.
Official reproduction code for 'Efficient LLM Preference Classification Through Position Bias Mitigation and Architectural Symmetry' (Applied Intelligence APIN-D-26-05359)
Counterfactual learning-to-rank for marketplace search logs in PySpark: position-bias estimation, IPS-weighted training, NDCG evaluation against known ground truth
Open-source implementation of the typed-decision pattern popularised by TypeSafe's Jev: read a decision out of a small language model's logits, in the browser. Library, benchmarks and paper. Not affiliated with TypeSafe.
Monthly bias audits of LLM judges
Ranking evaluation with error bars: NDCG, MRR and MAP with confidence intervals, plus position-bias correction for click logs. No dependencies.
To associate your repository with the position-bias topic, visit your repo's landing page and select "manage topics."