Skip to content
← Back to explorer

HFEPX Hub

CS.LG + Pairwise Preference Papers

Updated from current HFEPX corpus (Feb 27, 2026). 21 papers are grouped in this hub page. Common evaluation modes: Automatic Metrics, Human Eval. Most common rater population: Domain Experts. Common annotation unit: Pairwise. Frequent quality control: Calibration. Frequently cited benchmark: LiveCodeBench. Common metric signal: accuracy. Use this page to compare protocol setup, judge behavior, and labeling design decisions before running new eval experiments. Newest paper in this set is from Feb 26, 2026.

Papers: 21 Last published: Feb 26, 2026 Global RSS Tag RSS
Cs.LGPairwise Preference

Research Narrative

Grounded narrative Model: deterministic-grounded Source: persisted

Updated from current HFEPX corpus (Feb 27, 2026). This page tracks 21 papers for CS.LG + Pairwise Preference Papers. Dominant protocol signals include automatic metrics, human evaluation, simulation environments, with frequent benchmark focus on LiveCodeBench, Mathbench and metric focus on accuracy, agreement. Use the grounded sections below to prioritize reproducible protocol choices, benchmark-matched comparisons, and judge-vs-human evaluation checks.

Why This Matters For Eval Research

Protocol Takeaways

Benchmark Interpretation

  • LiveCodeBench appears in 4.8% of hub papers (1/21); use this cohort for benchmark-matched comparisons.
  • Mathbench appears in 4.8% of hub papers (1/21); use this cohort for benchmark-matched comparisons.

Metric Interpretation

  • accuracy is reported in 9.5% of hub papers (2/21); compare with a secondary metric before ranking methods.
  • agreement is reported in 4.8% of hub papers (1/21); compare with a secondary metric before ranking methods.

Researcher Checklist

  • Maintain strength on Papers with explicit human feedback. Coverage is strong (100% vs 45% target).
  • Close gap on Papers reporting quality controls. Coverage is a replication risk (4.8% vs 30% target).
  • Close gap on Papers naming benchmarks/datasets. Coverage is a replication risk (9.5% vs 35% target).
  • Tighten coverage on Papers naming evaluation metrics. Coverage is usable but incomplete (23.8% vs 35% target).
  • Close gap on Papers with known rater population. Coverage is a replication risk (14.3% vs 35% target).
  • Maintain strength on Papers with known annotation unit. Coverage is strong (52.4% vs 35% target).

Papers with explicit human feedback

Coverage is strong (100% vs 45% target).

Papers reporting quality controls

Coverage is a replication risk (4.8% vs 30% target).

Papers naming benchmarks/datasets

Coverage is a replication risk (9.5% vs 35% target).

Papers naming evaluation metrics

Coverage is usable but incomplete (23.8% vs 35% target).

Papers with known rater population

Coverage is a replication risk (14.3% vs 35% target).

Papers with known annotation unit

Coverage is strong (52.4% vs 35% target).

Suggested Reading Order

  1. 1. Moral Preferences of LLMs Under Directed Contextual Influence

    Start here for detailed protocol reporting, including rater and quality-control evidence.

  2. 2. Duel-Evolve: Reward-Free Test-Time Scaling via LLM Self-Preferences

    Start here for detailed protocol reporting, including rater and quality-control evidence.

  3. 3. Probing Graph Neural Network Activation Patterns Through Graph Topology

    Start here for detailed protocol reporting, including rater and quality-control evidence.

  4. 4. Incentivizing Agentic Reasoning in LLM Judges via Tool-Integrated Reinforcement Learning

    Include a human-eval paper to anchor calibration against automated judge settings.

  5. 5. Hierarchical Reward Design from Language: Enhancing Alignment of Agent Behavior with Human Specifications

    Adds automatic metrics with pairwise preferences for broader coverage within this hub.

  6. 6. Simplifying Outcomes of Language Model Component Analyses with ELIA

    Adds automatic metrics with pairwise preferences for broader coverage within this hub.

  7. 7. Align Once, Benefit Multilingually: Enforcing Multilingual Consistency for LLM Safety Alignment

    Adds automatic metrics with pairwise preferences for broader coverage within this hub.

  8. 8. Who can we trust? LLM-as-a-jury for Comparative Assessment

    Adds automatic metrics with pairwise preferences for broader coverage within this hub.

Known Limitations

  • Only 4.8% of papers report quality controls; prioritize calibration/adjudication evidence.
  • Rater population is under-specified (14.3% coverage).
  • Narrative synthesis is grounded in metadata and abstracts only; full-paper implementation details are not parsed.

Research Utility Links

human_eval vs automatic_metrics

both=0, left_only=1, right_only=19

0 papers use both Human Eval and Automatic Metrics.

automatic_metrics vs simulation_env

both=0, left_only=19, right_only=1

0 papers use both Automatic Metrics and Simulation Env.

human_eval vs simulation_env

both=0, left_only=1, right_only=1

0 papers use both Human Eval and Simulation Env.

Benchmark Brief

LiveCodeBench

Coverage: 1 papers (4.8%)

1 papers (4.8%) mention LiveCodeBench.

Examples: Duel-Evolve: Reward-Free Test-Time Scaling via LLM Self-Preferences

Benchmark Brief

Mathbench

Coverage: 1 papers (4.8%)

1 papers (4.8%) mention Mathbench.

Examples: Duel-Evolve: Reward-Free Test-Time Scaling via LLM Self-Preferences

Benchmark Brief

Retrieval

Coverage: 1 papers (4.8%)

1 papers (4.8%) mention Retrieval.

Examples: Distributional Vision-Language Alignment by Cauchy-Schwarz Divergence

Metric Brief

accuracy

Coverage: 2 papers (9.5%)

2 papers (9.5%) mention accuracy.

Examples: Duel-Evolve: Reward-Free Test-Time Scaling via LLM Self-Preferences , Precise Attribute Intensity Control in Large Language Models via Targeted Representation Editing

Metric Brief

agreement

Coverage: 1 papers (4.8%)

1 papers (4.8%) mention agreement.

Examples: Multi-Objective Alignment of Language Models for Personalized Psychotherapy

Metric Brief

calibration

Coverage: 1 papers (4.8%)

1 papers (4.8%) mention calibration.

Examples: Who can we trust? LLM-as-a-jury for Comparative Assessment

Most Cited In This Hub

Fast path to methods with the strongest citation traction in this scope.

Papers: Moral Preferences of LLMs Under Directed Contextual Influence , Duel-Evolve: Reward-Free Test-Time Scaling via LLM Self-Preferences , Probing Graph Neural Network Activation Patterns Through Graph Topology

Most Recent

Fast path to latest protocol changes and newly published evaluation setups.

Papers: Moral Preferences of LLMs Under Directed Contextual Influence , Duel-Evolve: Reward-Free Test-Time Scaling via LLM Self-Preferences , Probing Graph Neural Network Activation Patterns Through Graph Topology

Best Protocol Detail

Papers with explicit rater/unit metadata and quality-control signals for reproducibility.

Papers: Moral Preferences of LLMs Under Directed Contextual Influence , Duel-Evolve: Reward-Free Test-Time Scaling via LLM Self-Preferences , Probing Graph Neural Network Activation Patterns Through Graph Topology

Top Papers

Related Hubs