News without the noise
57 results for “ALFWorld” · page 3 of 3
Research · Agents
RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning
RetireOPD uses a skill-conditioned teacher for dense supervision, then automatically retires it when the student is ready to continue with reinforcement learning alone.
Zhejiang University, Alibaba GroupCode
Research · Agents
MileGPO: Milestone Inference with Local Evidence for Graph-Based Policy Optimization of Long-Horizon LLM Agents
MileGPO extracts intermediate rewards from existing rollouts by identifying milestones and traps, weighting their reliability, and comparing alternative actions from the same state.
Beijing Jiaotong University, Peking University
Research · Reinforcement learning
From Outcomes to Steps: Reliable Success-Aware Critic Learning for Long-Horizon Language Agents
SCALE adds an explicit success probability head to actor-critic RL and uses reliability-weighted changes in that probability to credit intermediate actions, raising success rates on five agent benchmarks.
ALFWorld: 97.4% success rate (as reported)
Shanghai Jiao Tong University, Huawei Noah's Ark Lab
Research · Agents
Co-Evolving Acting and Forecasting for Language Agents
CoPE trains agents to predict several subsequent actions from each interaction history, improving five benchmark results without adding inference-time computation.
ALFWorld: 95.5% success rate (as reported)
University of Illinois Urbana-Champaign, University of California San Diego, Meta AI
Research · Agents
Multi-Agent Transactive Memory
The authors propose Multi-Agent Transactive Memory, a repository for agent-generated trajectories that improves task success and reduces steps in interactive environments.
Carnegie Mellon University, University of California, Berkeley
Research · Reinforcement learning
TGRL: Temperature-Grouped Reinforcement Learning for Efficient Exploration in LLMs
By contrasting rollouts at different temperatures, TGRL allocates an explicit exploration signal without increasing sample budgets.
ALFWorld: 86.7% Success Rate (as reported)
University of Chinese Academy of Sciences, Meituan, MAIS&NLPR, Institute of Automation, Chinese Academy of SciencesCode
Research · Agents
Teacher-Student Gaps Are Not Enough: Outcome-Guided On-Policy Distillation for Multi-Turn Autonomous Agents
Outcome-guided weighting identifies which teacher interventions actually help the current student complete multi-turn tasks.
ALFWorld: 18 Rounds (as reported)
Qwen Large Model Application Team, Alibaba, Peking University, University of Chinese Academy of Sciences
Research · Agents
RSI-Router: Evolving Subtask-Level LLM Routing and Skills for Cost-Efficient Agents
RSI-Router learns when to assign subtasks to large or small language models, along with reusable model-specific execution guidance.
ALFWorld: 98.44% success rate (as reported)
Shanghai Artificial Intelligence Laboratory
Research · Reinforcement learning
Targeting Pivotal Decisions for Credit Assignment in Agentic Reinforcement Learning
ProVer uses a model to identify consequential action segments, then validates their value through sampled outcomes before assigning extra training credit.
ALFWorld: 90.8% task success rate (as reported)
University of California, Davis, Microsoft, University of Washington
Research · Agents
Does Learning to Predict the World Help Agents Act? Auditing World-Model Post-Training
Controlled experiments found that mismatched observations and random rewards could improve agents despite providing no correct information about environment transitions.
ALFWorld: 56.2% pass@64 (as reported)
Xi’an Jiaotong University, University of Southern California, University of the Chinese Academy of Sciences
Research · Agents
Beyond Memory: Harnessing Long-Horizon Agents with Explicit Belief States
PoS continually revises an agent’s view of the current world, checks it for inconsistencies, and intervenes when progress stalls.
ALFWorld: 97.01% task success (as reported)
Nankai University, Alibaba Group, Tsinghua University
Research · Agents
Retrieval-Augmented Skill Optimization via Cross-Harness Adaptation
RASO adapts procedural knowledge from an external skill corpus, using it both to initialize agent instructions and repair them after execution failures.
ALFWorld: 51 performance (as reported)
Korea University, KAIST, Meta AI
Research · Agents
UpliftMem: Learning Set-Level Uplift for Agent Memory Retrieval
UpliftMem learns which sets of stored examples improve task outcomes, then directs limited training rollouts toward alternatives most likely to change retrieval decisions.
ALFWorld: 70.9% success rate (as reported)
MIIT Key Laboratory of Data and Decision Intelligence, Beihang University
Research · Reinforcement learning
KV-streams for Efficient Compaction in Agentic Reinforcement Learning
Across text games and software engineering tasks, KV-streams cut training time while preserving the performance of conventional re-prefill compaction.
ALFWorld: 3.1 wall-clock speedup (as reported)
Mila, Microsoft, McGill University
Research · Reinforcement learning
Mitigating the Length-Scaling Tax with Online Distillation
Length Self-Distillation reduces excess response length on easy queries without sacrificing accuracy on hard ones.
BrowseComp-Plus: 31.65% Pass@1 (as reported)
ByteDance Seed, Tongji University, Peking University
Research · Agents
Dense Is Not Enough: Hierarchical Supervision Allocation for Long-Horizon On-Policy Distillation
The LENS-OPD method outperforms dense token-level supervision by dynamically selecting which student behaviors to correct based on future impact and learnability.
ALFWorld: 46.3% SR (as reported)
Ant Group, Alibaba International Digital Commerce Group, University of Science and Technology of China
Research digest
Across routing, skill retrieval, harness optimization, and credit assignment, targeted structure outperformed brute-force evaluation.