SENTINEL generates targeted tasks from model failures in a Controller-Proposer-Solver loop, raising Pass^1 from 66.4 to 74.9 on Tau2-Bench Retail and outperforming standard RL.
Self-challenging language model agents
10 Pith papers cite this work. Polarity classification is still indexing.
10
Pith papers citing it
citation-role summary
background 3
citation-polarity summary
years
2026 10roles
background 3polarities
background 3representative citing papers
An agentic weak–strong Self-Instruct loop, optionally meta-optimized, produces synthetic data that trains small models better than standard CoT Self-Instruct across three domains.
PROTON is a test-time online OOD detection module for medical VLMs that maintains a prototype bank from confident predictions and adaptively combines prototype distance with MCM scoring via variance statistics, improving AUROC across covariate, semantic, and far-OOD shifts on FLAIR+FIVES without tra
BenchEvolver evolves coding problem solutions to generate harder, valid tasks, producing LiveCodeBench-Plus where frontier models score 27.5-62.6% and enabling RL gains on held-out tests.
unix-ctf procedurally generates 656 Unix CTF tasks across 155 techniques; fine-tuning Qwen3-8B on them raises solve rate from 11.6% to 43.6% on a 15-skill holdout and yields +33 pp in Forensics on InterCode-CTF.
G-Zero uses the Hint-δ intrinsic reward to drive co-evolution between a Proposer and Generator via GRPO and DPO, providing a theoretical suboptimality guarantee for self-improvement from internal dynamics alone.
POP bootstraps post-training signals for open-ended LLM tasks by synthesizing rubrics during self-play on pretraining corpus, yielding performance gains on Qwen-2.5-7B across healthcare QA, creative writing, and instruction following.
LLM agents trained with a task-success reward on self-generated knowledge can spontaneously explore and adapt to new environments without any rewards or instructions at inference, yielding 20% gains on web tasks and allowing a 14B model to beat Gemini-2.5-Flash.
Vigil deploys a proactive agent for full on-call lifecycle support with autonomous self-improvement from human-resolved cases.
A three-stage training pipeline internalizes world-model simulation and success estimation in LLM agents for improved planning on search and math tasks.
citing papers explorer
-
SENTINEL: Failure-Driven Reinforcement Learning for Training Tool-Using Language Model Agents
SENTINEL generates targeted tasks from model failures in a Controller-Proposer-Solver loop, raising Pass^1 from 66.4 to 74.9 on Tau2-Bench Retail and outperforming standard RL.
-
Autodata: An agentic data scientist to create high quality synthetic data
An agentic weak–strong Self-Instruct loop, optionally meta-optimized, produces synthetic data that trains small models better than standard CoT Self-Instruct across three domains.
-
PROTON: Prototype-Based Test-Time Online OOD Detection for Medical VLMs
PROTON is a test-time online OOD detection module for medical VLMs that maintains a prototype bank from confident predictions and adaptively combines prototype distance with MCM scoring via variance statistics, improving AUROC across covariate, semantic, and far-OOD shifts on FLAIR+FIVES without tra
-
BenchEvolver: Frontier Task Synthesis via Solution-Centric Evolution
BenchEvolver evolves coding problem solutions to generate harder, valid tasks, producing LiveCodeBench-Plus where frontier models score 27.5-62.6% and enabling RL gains on held-out tests.
-
unix-ctf: Procedural Environments for Unix-Competence Reinforcement Learning
unix-ctf procedurally generates 656 Unix CTF tasks across 155 techniques; fine-tuning Qwen3-8B on them raises solve rate from 11.6% to 43.6% on a 15-skill holdout and yields +33 pp in Forensics on InterCode-CTF.
-
G-Zero: Self-Play for Open-Ended Generation from Zero Data
G-Zero uses the Hint-δ intrinsic reward to drive co-evolution between a Proposer and Generator via GRPO and DPO, providing a theoretical suboptimality guarantee for self-improvement from internal dynamics alone.
-
Bootstrapping Post-training Signals for Open-ended Tasks via Rubric-based Self-play on Pre-training Text
POP bootstraps post-training signals for open-ended LLM tasks by synthesizing rubrics during self-play on pretraining corpus, yielding performance gains on Qwen-2.5-7B across healthcare QA, creative writing, and instruction following.
-
Training LLM Agents for Spontaneous, Reward-Free Self-Evolution via World Knowledge Exploration
LLM agents trained with a task-success reward on self-generated knowledge can spontaneously explore and adapt to new environments without any rewards or instructions at inference, yielding 20% gains on web tasks and allowing a 14B model to beat Gemini-2.5-Flash.
-
Help Without Being Asked: A Deployed Proactive Agent System for On-Call Support with Continuous Self-Improvement
Vigil deploys a proactive agent for full on-call lifecycle support with autonomous self-improvement from human-resolved cases.
-
Internalizing the Future: A Unified Agentic Training Paradigm for World Model Planning
A three-stage training pipeline internalizes world-model simulation and success estimation in LLM agents for improved planning on search and math tasks.