Skip to content
← Back to explorer

HFEPX Hub

Law Papers

Updated from current HFEPX corpus (Apr 12, 2026). 42 papers are grouped in this hub page.

Read Full Context

Updated from current HFEPX corpus (Apr 12, 2026). 42 papers are grouped in this hub page. Common evaluation modes: Automatic Metrics, Simulation Env. Most common rater population: Domain Experts. Common annotation unit: Multi Dim Rubric. Frequent quality control: Adjudication. Frequently cited benchmark: GSM8K. 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 Mar 9, 2026.

Papers: 42 Last published: Mar 9, 2026 Global RSS Tag RSS
Law

Researcher Quick Triage

This hub is best used for protocol triage and replication planning from abstract-level evidence. Quality band: Medium .

High-Signal Coverage

100.0%

42 / 42 sampled papers are not low-signal flagged.

Replication-Ready Set

8

Benchmark + metric + eval mode explicitly present.

Judge/Human Comparability

0

Papers containing both `human_eval` and `llm_as_judge`.

  • 8 papers are replication-ready (benchmark + metric + explicit evaluation mode).
  • 0 papers support judge-vs-human agreement analysis.
  • 3 papers report explicit quality controls (calibration/adjudication/IAA).

Primary action: Start with the top 2 papers in “Start Here”, then validate assumptions in the protocol matrix.

Currently showing only replication-ready papers in ranking and matrix sections (8 papers).

Need evaluators for this research workflow?

Post a Job →

Why This Matters For Eval Research

  • 54.8% of papers report explicit human-feedback signals, led by pairwise preferences.
  • automatic metrics appears in 61.9% of papers in this hub.
  • GSM8K is a recurring benchmark anchor for cross-paper comparisons in this page.

Protocol Takeaways

  • Most common quality-control signal is adjudication (2.4% of papers).
  • Rater context is mostly domain experts, and annotation is commonly multi-dimensional rubrics; use this to scope replication staffing.
  • Compare papers that report both human_eval and llm_as_judge to quantify judge-human agreement drift.

Benchmark Interpretation

  • GSM8K appears in 4.8% of hub papers (2/42); use this cohort for benchmark-matched comparisons.
  • HLE appears in 4.8% of hub papers (2/42); use this cohort for benchmark-matched comparisons.

Metric Interpretation

  • accuracy is reported in 31% of hub papers (13/42); compare with a secondary metric before ranking methods.
  • cost is reported in 16.7% of hub papers (7/42); compare with a secondary metric before ranking methods.
Researcher Checklist (Expanded)

Researcher Checklist

  • Strong: Papers with explicit human feedback

    Coverage is strong (54.8% vs 45% target).

  • Gap: Papers reporting quality controls

    Coverage is a replication risk (7.1% vs 30% target).

  • Moderate: Papers naming benchmarks/datasets

    Coverage is usable but incomplete (26.2% vs 35% target).

  • Strong: Papers naming evaluation metrics

    Coverage is strong (59.5% vs 35% target).

  • Moderate: Papers with known rater population

    Coverage is usable but incomplete (28.6% vs 35% target).

  • Moderate: Papers with known annotation unit

    Coverage is usable but incomplete (31% vs 35% target).

Strengths

  • Strong human-feedback signal (54.8% of papers).
  • Contains both human-eval and LLM-as-judge protocols for head-to-head methodology comparison.
  • Agentic evaluation appears in 54.8% of papers.

Known Gaps

  • Only 7.1% of papers report quality controls; prioritize calibration/adjudication evidence.
  • LLM-as-judge appears without enough inter-annotator agreement reporting.

Suggested Next Analyses

  • Compare papers that report both human_eval and llm_as_judge to quantify judge-human agreement drift.
  • Stratify by benchmark (GSM8K vs HLE) before comparing methods.
  • Track metric sensitivity by reporting both accuracy and cost.
  • Add inter-annotator agreement checks when reproducing these protocols.
Recommended Queries (Expanded)

Recommended Queries

Start with These 3

Use these when you need one protocol anchor, one benchmark anchor, and one recent comparison point before reading the wider hub.

Start Here (Best First 6)

Ranked for protocol completeness (human signal, benchmark + metric anchors, quality controls, and judge/human overlap).

Protocol Matrix (Top 12)

Use this to quickly compare protocol ingredients instead of scanning long prose.

Protocol Diff (Top Papers)

Fast side-by-side comparison for the highest-ranked papers in this hub.

Signal HLE-Verified: A Systematic Verification and Structu… Xpertbench: Expert Level Tasks with Rubrics-Based E… Stabilizing Iterative Self-Training with Verified R…
Human Feedback Expert Verification, Critique EditRubric Rating, Expert VerificationPairwise Preference
Evaluation Modes Automatic MetricsAutomatic MetricsAutomatic Metrics
Benchmarks HLEXpertbenchGSM8K
Metrics AccuracySuccess rateAccuracy
Quality Controls AdjudicationNot reportedNot reported
Rater Population Domain ExpertsDomain ExpertsUnknown
Annotation Unit UnknownMulti Dim RubricUnknown
Suggested Reading Order (Extended)

This section is intentionally expanded only when needed; use “Start Here” above for a faster pass.

Suggested Reading Order

  1. Strategic Persuasion with Trait-Conditioned Multi-Agent Systems for Iterative Legal Argumentation

    Start here for detailed protocol reporting and quality-control evidence. Signals: simulation environments. Abstract: Strategic interaction in adversarial domains such as law, diplomacy, and negotiation is mediated by language,.

  2. TriAttention: Efficient Long Reasoning with Trigonometric KV Compression

    Start here for detailed protocol reporting and quality-control evidence. Signals: automatic metrics + pairwise preferences. Focus: accuracy. Abstract: Via the trigonometric series, we use the distance preference characterized.

  3. Unifying Group-Relative and Self-Distillation Policy Optimization via Sample Routing

    Start here for detailed protocol reporting and quality-control evidence. Signals: automatic metrics. Focus: cost. Abstract: Reinforcement learning with verifiable rewards (RLVR) has become a standard paradigm for post-training.

  4. \$OneMillion-Bench: How Far are Language Agents from Human Experts?

    Include a human-eval paper to calibrate against judge-based evaluation settings. Signals: automatic metrics + rubric ratings. Focus: Onemillion-Bench / accuracy. Abstract: We adopt a rubric-based evaluation protocol scoring.

  5. HLE-Verified: A Systematic Verification and Structured Revision of Humanity's Last Exam

    Include a human-eval paper to calibrate against judge-based evaluation settings. Signals: automatic metrics + expert verification. Focus: HLE / accuracy. Abstract: Overall, HLE-Verified improves HLE-style evaluations by reducing.

  6. LEXam: Benchmarking Legal Reasoning on 340 Law Exams

    Include an LLM-as-judge paper to test judge design and agreement assumptions. Signals: LLM-as-judge. Focus: accuracy. Abstract: Deploying an ensemble LLM-as-a-Judge paradigm with rigorous human expert validation, we demonstrate.

  7. APEX-Agents

    Adds automatic metrics with rubric ratings for broader protocol coverage within this hub. Signals: automatic metrics + rubric ratings. Focus: pass@1. Abstract: We open source the APEX-Agents benchmark.

  8. Xpertbench: Expert Level Tasks with Rubrics-Based Evaluation

    Adds automatic metrics with rubric ratings for broader protocol coverage within this hub. Signals: automatic metrics + rubric ratings. Focus: Xpertbench / success rate. Abstract: Each task uses.

Known Limitations

Known Limitations

  • Only 7.1% of papers report quality controls; prioritize calibration/adjudication evidence.
  • LLM-as-judge appears without enough inter-annotator agreement reporting.
  • Narrative synthesis is grounded in metadata and abstracts only; full-paper implementation details are not parsed.
Research Utility Snapshot (Detailed)

Research Utility Snapshot

Human Feedback Mix

  • Pairwise Preference (11)
  • Expert Verification (7)
  • Rubric Rating (5)
  • Critique Edit (2)

Evaluation Modes

  • Automatic Metrics (26)
  • Simulation Env (4)
  • Human Eval (2)
  • Llm As Judge (2)

Top Benchmarks

  • GSM8K (2)
  • HLE (2)
  • AdvBench (1)
  • Cow Bench (1)

Top Metrics

  • Accuracy (13)
  • Cost (7)
  • Coherence (2)
  • Jailbreak success rate (2)

Rater Population Mix

  • Domain Experts (12)

Quality Controls

  • Adjudication (1)
  • Calibration (1)
  • Gold Questions (1)
Coverage diagnostics (sample-based): human-feedback 54.8% · benchmarks 26.2% · metrics 59.5% · quality controls 7.1%.

Top Papers

Related Hubs

Need human evaluators for your AI research? Scale annotation with expert AI Trainers.