Your daily digest
Research papers, made simple to understand
The AI and machine learning papers people are reading most, each one rewritten in plain words: the problem, the idea, what was found and why it matters.
- AI Agents
Trajectory2Gym improves tool-use agents by synthesizing and grounding compositional interactions
Trajectory2Gym constructs training data focusing on compositional coverage and decision grounding, improving tool-use agent performance across five Qwen3 model scales by 16.9–29.8 points on three benchmarks, especially in multi-turn, stateful tasks.
Paper: Trajectory2Gym: Decision-Grounded Compositional Interaction Synthesis for Tool-Use Agents · Yihao Hu, Zhihao Wen, Yitong Lin and 5 others
338 upvotes on Hugging Face6 min readRead the article
- AI Agents
- Benchmarks & Evaluation
TraceTrial-Bench evaluates AI agent harness reliability separately from model performance
TraceTrial-Bench benchmarks AI agent harnesses by assessing seven reliability components through 95 tasks, separating agent task success from harness reliability, and enabling failure attribution across 4 harnesses, 9 models, and 3,420 runs.
Paper: Trial by Trace: Benchmarking AI Agent Harnesses for Reliability and Failure Attribution · Zhiyong Cao, Hengjie Zhu, Yongxin Wang and 20 others
268 upvotes on Hugging Face6 min readRead the article
- AI Agents
- Multimodal & Vision
StreamVibe enables proactive video agents to infer and adapt to evolving user intents
StreamVibe introduces a benchmark and models for proactive video agents that infer implicit and dynamic user intents from streaming video, improving response timing and quality by 10.7% and maintaining consistent interaction states over time.
Paper: StreamVibe: Proactive Video Agents for Evolving User Intent · Chengzhi Liu, Yuxiang Lai, Yuzhe Yang and 13 others
246 upvotes on Hugging Face6 min readRead the article
Browse earlier papers
2 days of papers in November 2026