Skip to content
← Back to explorer

HFEPX Hub

Demonstrations Or Red Team Papers

Updated from current HFEPX corpus (Apr 12, 2026). 130 papers are grouped in this hub page.

Read Full Context

Updated from current HFEPX corpus (Apr 12, 2026). 130 papers are grouped in this hub page. Common evaluation modes: Automatic Metrics, Simulation Env. Most common rater population: Domain Experts. Common annotation unit: Trajectory. Frequent quality control: Calibration. Frequently cited benchmark: AdvBench. 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 22, 2026.

Papers: 130 Last published: Mar 22, 2026 Global RSS Tag RSS
DemonstrationsRed Team

Researcher Quick Triage

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

Analysis blocks below are computed from the currently loaded sample (60 of 130 total papers in this hub).

High-Signal Coverage

100.0%

60 / 60 sampled papers are not low-signal flagged.

Replication-Ready Set

8

Benchmark + metric + eval mode explicitly present.

Judge/Human Comparability

1

Papers containing both `human_eval` and `llm_as_judge`.

  • 8 papers are replication-ready (benchmark + metric + explicit evaluation mode).
  • 1 papers support judge-vs-human agreement analysis.
  • 1 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.

Need evaluators for this research workflow?

Post a Job →

Why This Matters For Eval Research

  • 100% of papers report explicit human-feedback signals, led by demonstration data.
  • automatic metrics appears in 25.4% of papers in this hub.
  • AdvBench is a recurring benchmark anchor for cross-paper comparisons in this page.

Protocol Takeaways

  • 1 sampled papers report both human evaluation and LLM-as-judge, supporting direct agreement checks.
  • Most common quality-control signal is rater calibration (0.8% of papers).
  • Rater context is mostly domain experts, and annotation is commonly trajectory-level annotation; use this to scope replication staffing.

Benchmark Interpretation

  • AdvBench appears in 1.5% of hub papers (2/130); use this cohort for benchmark-matched comparisons.
  • DROP appears in 1.5% of hub papers (2/130); use this cohort for benchmark-matched comparisons.

Metric Interpretation

  • accuracy is reported in 6.9% of hub papers (9/130); compare with a secondary metric before ranking methods.
  • jailbreak success rate is reported in 6.9% of hub papers (9/130); compare with a secondary metric before ranking methods.
Researcher Checklist (Expanded)

Researcher Checklist

  • Strong: Papers with explicit human feedback

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

  • Gap: Papers reporting quality controls

    Coverage is a replication risk (0.8% vs 30% target).

  • Gap: Papers naming benchmarks/datasets

    Coverage is a replication risk (16.2% vs 35% target).

  • Moderate: Papers naming evaluation metrics

    Coverage is usable but incomplete (30.8% vs 35% target).

  • Gap: Papers with known rater population

    Coverage is a replication risk (13.8% vs 35% target).

  • Gap: Papers with known annotation unit

    Coverage is a replication risk (13.1% vs 35% target).

Strengths

  • Strong human-feedback signal (100% of papers).
  • Contains both human-eval and LLM-as-judge protocols for head-to-head methodology comparison.

Known Gaps

  • Only 0.8% of papers report quality controls; prioritize calibration/adjudication evidence.
  • Rater population is under-specified (13.8% coverage).
  • Annotation unit is under-specified (13.1% coverage).

Suggested Next Analyses

  • Compare papers that report both human_eval and llm_as_judge to quantify judge-human agreement drift.
  • Stratify by benchmark (AdvBench vs DROP) before comparing methods.
  • Track metric sensitivity by reporting both accuracy and jailbreak success rate.
  • 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.

Paper HF Signal Eval Modes Benchmarks Metrics QC
AgentHER: Hindsight Experience Replay for LLM Agent Trajectory Relabeling

Mar 22, 2026

Yes Human Eval , Llm As Judge WebArena , ToolBench Precision , Pass@1 Not Reported
TraceSafe: A Systematic Assessment of LLM Guardrails on Multi-Step Tool-Calling Trajectories

Apr 8, 2026

Yes Automatic Metrics Tracesafe Bench Accuracy Not Reported
SemEval-2026 Task 6: CLARITY -- Unmasking Political Question Evasions

Mar 14, 2026

Yes Automatic Metrics Semeval F1 , F1 macro Not Reported
Jailbreak Foundry: From Papers to Runnable Attacks for Reproducible Benchmarking

Feb 27, 2026

Yes Llm As Judge AdvBench , Jbf Eval Success rate , Jailbreak success rate Not Reported
AJAR: Adaptive Jailbreak Architecture for Red-teaming

Jan 16, 2026

Yes Simulation Env Harmbench Success rate , Jailbreak success rate Not Reported
Meanings and Measurements: Multi-Agent Probabilistic Grounding for Vision-Language Navigation

Mar 19, 2026

Yes Simulation Env Mapg Bench Not Reported Not Reported
RedTeamCUA: Realistic Adversarial Testing of Computer-Use Agents in Hybrid Web-OS Environments

May 28, 2025

Yes Automatic Metrics Rtc Bench Jailbreak success rate Not Reported
A Simple and Efficient Jailbreak Method Exploiting LLMs' Helpfulness

Sep 17, 2025

Yes Automatic Metrics AdvBench Helpfulness Not Reported
Dyslexify: A Mechanistic Defense Against Typographic Attacks in CLIP

Aug 28, 2025

Yes Automatic Metrics DROP Accuracy Not Reported
A Coin Flip for Safety: LLM Judges Fail to Reliably Measure Adversarial Robustness

Feb 4, 2026

Yes Llm As Judge Reliablebench Not Reported Not Reported
Dual-Modality Multi-Stage Adversarial Safety Training: Robustifying Multimodal Web Agents Against Cross-Modal Attacks

Mar 4, 2026

Yes Simulation Env MiniWoB++ Not Reported Not Reported
IA2: Alignment with ICL Activations Improves Supervised Fine-Tuning

Sep 26, 2025

Yes Automatic Metrics Not Reported Accuracy , Cost Calibration

Protocol Diff (Top Papers)

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

Signal AgentHER: Hindsight Experience Replay for LLM Agent… TraceSafe: A Systematic Assessment of LLM Guardrail… SemEval-2026 Task 6: CLARITY -- Unmasking Political…
Human Feedback DemonstrationsRed TeamRed Team
Evaluation Modes Human Eval, Llm As JudgeAutomatic MetricsAutomatic Metrics
Benchmarks WebArena, ToolBenchTracesafe BenchSemeval
Metrics Precision, Pass@1AccuracyF1, F1 macro
Quality Controls Not reportedNot reportedNot reported
Rater Population UnknownUnknownDomain Experts
Annotation Unit TrajectoryTrajectoryUnknown
Suggested Reading Order (Extended)

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

Suggested Reading Order

  1. TraceSafe: A Systematic Assessment of LLM Guardrails on Multi-Step Tool-Calling Trajectories

    Start here for detailed protocol reporting and quality-control evidence. Signals: automatic metrics + red-team protocols. Focus: Tracesafe-Bench / accuracy. Abstract: As large language models (LLMs) evolve from static.

  2. State-of-the-Art Arabic Language Modeling with Sparse MoE Fine-Tuning and Chain-of-Thought Distillation

    Start here for detailed protocol reporting and quality-control evidence. Signals: automatic metrics + demonstration data. Focus: cost. Abstract: This paper introduces Arabic-DeepSeek-R1, an application-driven open-source Arabic LLM that.

  3. AgentHER: Hindsight Experience Replay for LLM Agent Trajectory Relabeling

    Include a human-eval paper to calibrate against judge-based evaluation settings. Signals: human evaluation + demonstration data. Focus: WebArena / precision. Abstract: AgentHER realises this idea through a four-stage.

  4. Jailbreak Foundry: From Papers to Runnable Attacks for Reproducible Benchmarking

    Include a human-eval paper to calibrate against judge-based evaluation settings. Signals: LLM-as-judge + red-team protocols. Focus: AdvBench / success rate. Abstract: This system enables a standardized AdvBench evaluation.

  5. VolleyBots: A Testbed for Multi-Drone Volleyball Game Combining Motion Control and Strategic Play

    Adds automatic metrics with demonstration data for broader protocol coverage within this hub. Signals: automatic metrics + demonstration data. Focus: win rate. Abstract: Robot sports, characterized by well-defined.

  6. Meanings and Measurements: Multi-Agent Probabilistic Grounding for Vision-Language Navigation

    Adds simulation environments with demonstration data for broader protocol coverage within this hub. Signals: simulation environments + demonstration data. Focus: Mapg-Bench. Abstract: Robots collaborating with humans must convert.

  7. AJAR: Adaptive Jailbreak Architecture for Red-teaming

    Adds simulation environments with red-team protocols for broader protocol coverage within this hub. Signals: simulation environments + red-team protocols. Focus: Harmbench / success rate. Abstract: Large language model.

Known Limitations

Known Limitations

  • Only 0.8% of papers report quality controls; prioritize calibration/adjudication evidence.
  • Rater population is under-specified (13.8% coverage).
  • 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

  • Demonstrations (69)
  • Red Team (61)
  • Pairwise Preference (7)
  • Rubric Rating (2)

Evaluation Modes

  • Automatic Metrics (33)
  • Simulation Env (16)
  • Llm As Judge (5)
  • Human Eval (1)

Top Benchmarks

  • AdvBench (2)
  • DROP (2)
  • HotpotQA (2)
  • Windowsagentarena (2)

Top Metrics

  • Accuracy (9)
  • Jailbreak success rate (9)
  • Success rate (9)
  • Cost (6)

Rater Population Mix

  • Domain Experts (16)
  • Mixed (2)

Quality Controls

  • Calibration (1)
Coverage diagnostics (sample-based): human-feedback 100.0% · benchmarks 26.7% · metrics 61.7% · quality controls 1.7%.

Top Papers

Related Hubs

Get Started

Join the #1 Platform for AI Training Talent

Where top AI builders and expert AI Trainers connect to build the future of AI.
Self-Service
Post a Job
Post your project and get a shortlist of qualified AI Trainers and Data Labelers. Hire and manage your team in the tools you already use.
Managed Service
For Large Projects
Done-for-You
We recruit, onboard, and manage a dedicated team inside your tools. End-to-end operations for large or complex projects.
For Freelancers
Join as an AI Trainer
Find AI training and data labeling projects across platforms, all in one place. One profile, one application process, more opportunities.