Searcharxiv⌕ Search

arXiv subjects

An-Sheng Jhang

Publications and source records attributed to An-Sheng Jhang.

4 recordsLinked to original sources

Multi-Estimate Elicitation Does Not Uniformly Debias Language Model Judgments

Multi-Candidate Verbalization (MCV) is a task- and model-dependent output protocol: repairing format does not guarantee better posterior accuracy, and requesting weights differs from reweighting an already generated candidate set. Across 31,672 planned/admitted evaluation units and 32,184 planned/admitted calls (excluding retries), we connect a frozen six-configuration cohort to exploratory scale/organization experiments and two independently preregistered post-cohort formal suites. Stage 1 retains 11 directional confirmations among 45 frozen tests; only increased post-outcome confirmation error on gpt-oss-120b exceeds its SESOI. P1 completes 1,920 terminal units: smaller response scales sharply lower operational string-match rates but increase Bayesian error. P2 completes 2,560 calls and 2,048 units: eliciting weights raises three errors by +2.75 / +2.05 / +3.88 pp, while within-generation weighting changes them by only -0.052 / -0.049 / -0.067 pp, a 42x-58x magnitude difference. Unconstrained joint generation with post-hoc normalization offers a local normalized-error improvement for three exhaustive options, without confirmed accuracy improvement. These results distinguish format validity, numerical accuracy and psychological susceptibility; a descriptive three-class behavioral spectrum organizes heterogeneity without identifying internal causes. Materials: https://github.com/asj252/llm-judgment-studies

cs.CL↗

From Imperfective Paradox to Moorean Absurdity: Single-Chain Reasoning Lock-In and Multi-Candidate Verbalization Across Aspectual and Pragmatic Ambiguity

Natural language entailment and coherence judgments often depend on whether an event merely occurred, reached a telic culmination, satisfied a plan, or was evaluated under multiple pragmatic standards. We test observable consequences of the single-chain premature lock-in hypothesis across four preregistered, sealed formal experiments on Qwen3-8B-Q8_0 (16,640 formal units, temperature=0.0): 15 of 16 hypothesis slots retain confirmatory status (12 rejecting H0 under within-experiment Holm control at alpha_fam=0.05; E3_H3 demoted to descriptive status). In Experiment 1, Direct-CoT vs. MCU differs by 25.00 pp on telic high-confidence completion (Holm p=1.31239e-05), while NLI-CoT collapses by -91.25 pp on uninterrupted atelic items (Holm p=1.81805e-21). Multi-Candidate Verbalization (MCV) elicits K=5 weighted estimates in one call, separating the prompt from unweighted (MCU) and self-weighted (MCW) readouts. Mandarin (E2) and English (E3) expose an Interruption-Anchored Realis Paradox under NLI-CoT. In Experiment 4, MCU-CF raises perceived coherence on implicit Moorean sentences by 26.64 pp over Direct-CoT (Holm p=0.0006), alongside Direct-CF - Direct-CoT = 29.15 pp and MCU-CoT - Direct-CoT = 17.31 pp. A 1,800-call OpenAI ablation and a stage="dev" P16.7 cross-model extension (4,240 planned/collected units across five external models: 4,222 valid, 18 format-missing, 0 transport-missing; <2% missing per slice; OpenAI gpt-4o-mini at top_logprobs=20 achieves 1,000/1,000 valid units) show that discrete NLI-CoT and scalar Direct-CoT judgments dissociate across models. Confirmatory effects concern observable prompt-and-readout protocols; internal premature lock-in remains a hypothesis. All code, stimuli, preregistered analysis plans, and verification tools are available at https://github.com/asj252/llm-judgment-studies.

cs.CL↗

Hearsay or Fact? How Framing and Elicitation Shape LLM Probability Updates

We measure LLM probability updates on $δ$-SNLI premise-hypothesis pairs across the Multi-Candidate Verbalization (MCV) family. Separate calls elicit priors and posteriors; directional movement s signs changes by the annotated direction. We cross stated vs. reported wording with Direct single estimates (D0) vs. Multi-Candidate Weighted elicitation (MCW; five weighted estimates, V0, weighted mean M_W). Frozen formal tests cover 15 configurations (23,400 planned units): 3 primary (Qwen3-8B thinking off, OLMo-2-7B, gpt-6.1-sol via codex; 6,600), 4 CLI-harnessed configurations (gpt-oss-120b, claude-sonnet-5.5, gemini-3.8-flash low/high; dnli-agy; 5,600), and 8 un-harnessed REST API configurations (dnli-api; 11,200). We add a 20-configuration, 8-provider dev ladder (3B-550B; 6,000 units) and a 2-model OpenAI 1+1+4 ablation of D0, Direct-CoT (D_CoT), MCW (V0), and Multi-Candidate Unweighted (MCU; V2 / M_U; 1,800 calls). All 15 reject H1 (s = +16.1 to +28.9 points) and H4a (endpoint reduction +0.065 to +0.391); 13 reject H2 (+2.0 to +8.9 points), except gpt-oss-120b (+0.9) and sf:qwen3.5-4b (-0.64). In 14 of 15 formal configurations, readouts obey Top-1 > MCW (M_W) > MCU (M_U). On 120 common pairs, API Qwen3 tracks local Qwen3 (H1 +18.72 vs. +18.90; H4a +0.265 vs. +0.254; H2 +3.78 vs. +1.67; H4b -9.03 vs. -6.70); local H2 rejects on all 240 pairs (+1.97), not this subset. API Gemma 31B (formal H4b -2.85, Top-1 +0.64; spread -13.3% / +7.6%) and Groq gpt-oss-120b (dev -4.15, +0.40; -18.4% / +12.2%) show small MCW costs and stated narrowing/reported widening without CLI harnesses. In OpenAI ablations, D_CoT keeps numeric-token entropy <1 bit; MCU (V2) shows smaller movement loss but halves spread and $ΔH_2$ vs. MCW (V0). Because $δ$-NLI annotates only direction, we make no claim about appropriate update magnitude. Code/data: https://github.com/asj252/llm-judgment-studies

cs.CL↗

Analytical structure, dynamics, and coarse-graining of a kinetic model of an active fluid

We analyze one of the simplest active suspensions with complex dynamics: a suspension of immotile "Extensor" particles that exert active extensile dipolar stresses on the fluid in which they are immersed. This is relevant to several experimental systems, such as recently studied tripartite rods that create extensile flows by consuming a chemical fuel. We first describe the system through a Doi-Onsager kinetic theory based on microscopic modeling. This theory captures the active stresses produced by the particles that can drive hydrodynamic instabilities, as well as the steric interactions of rod-like particles that lead to nematic alignment. This active nematic system yields complex flows and disclination defect dynamics very similar to phenomenological Landau-deGennes Q-tensor theories for active nematic fluids, as well as by more complex Doi-Onsager theories for polar microtubule/motor-protein systems. We apply the quasi-equilibrium Bingham closure, used to study suspensions of passive microscopic rods, to develop a non-standard Q-tensor theory. We demonstrate through simulation that this "BQ-tensor" theory gives an excellent analytical and statistical accounting of the suspension's complex dynamics, at a far reduced computational cost. Finally, we apply the BQ-tensor model to study the dynamics of Extensor suspensions in circular and bi-concave domains. In circular domains, we reproduce previous results for systems with weak nematic alignment, but for strong alignment find novel dynamics with activity-controlled defect production and absorption at the boundaries of the domain. In bi-concave domains, a Fredericks-like transition occurs as the width of the neck connecting the two disks is varied.

cond-mat.soft↗