SkillTTA synthesizes temporary task-specific skills from retrieved training trajectories to boost LLM agent Pass@1 scores on SpreadsheetBench and BigCodeBench without parameter updates.
hub
arXiv preprint arXiv:2508.05668 , year=
14 Pith papers cite this work. Polarity classification is still indexing.
14
Pith papers citing it
hub tools
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
DeepRefine refines agent-compiled knowledge bases via multi-turn abductive diagnosis and RL training with a GBD reward, yielding consistent downstream task gains.
ToolPRM provides fine-grained intra-call process supervision via a new dataset and reward model, outperforming outcome and coarse-grained alternatives on function-calling benchmarks.
Arbor combines a coordinator, executors, and a hypothesis tree to enable cumulative autonomous research, outperforming Codex and Claude Code by over 2.5x on six real tasks and reaching 86.36% Any Medal on MLE-Bench Lite.
MaterEval generates paired informed and blind evaluations as preference signals to improve small open-source LLMs on high-entropy alloy assessment, approaching closed-source performance without external retrieval.
Malicious actors could use AI agents to submit large numbers of fake papers, inflating the submission count and thereby raising the acceptance odds for a small set of chosen legitimate papers under stable conference acceptance rates.
LMM-Searcher uses file-based visual UIDs and a fetch tool plus 12K synthesized trajectories to fine-tune a multimodal agent that scales to 100-turn horizons and reaches SOTA among open-source models on MM-BrowseComp and MMSearch-Plus.
Retrievers trained on agent trajectories via the LRAT framework improve evidence recall, task success, and efficiency in agentic search benchmarks.
MultiSearch uses parallel multi-query retrieval plus explicit merging inside a reinforcement-learning loop to improve retrieval-augmented reasoning, outperforming baselines on seven QA benchmarks.
SiriusHelper deploys an LLM agent with intent routing, DeepSearch multi-hop retrieval, and automated SOP distillation to outperform alternatives and reduce ticket volume by 20.8% on Tencent's big data platform.
LLM agent progress depends on externalizing cognitive functions into memory, skills, protocols, and harness engineering that coordinates them reliably.
ERL trains LLMs to erase faulty reasoning steps and regenerate them in place, yielding gains of up to 8.48% EM on multi-hop QA benchmarks like HotpotQA.
A deep research agent incorporates progressive confidence estimation and calibration to produce trustworthy reports with transparent confidence scores on claims.
The paper reviews conceptual foundations, methodological innovations, effective designs, critical challenges, and future directions for LLM-based Agentic Reinforcement Learning.
citing papers explorer
-
Towards Long-horizon Agentic Multimodal Search
LMM-Searcher uses file-based visual UIDs and a fetch tool plus 12K synthesized trajectories to fine-tune a multimodal agent that scales to 100-turn horizons and reaches SOTA among open-source models on MM-BrowseComp and MMSearch-Plus.
-
Rethinking Agentic Reinforcement Learning In Large Language Models
The paper reviews conceptual foundations, methodological innovations, effective designs, critical challenges, and future directions for LLM-based Agentic Reinforcement Learning.