AI Agent Papers2026
Proposes SkillOpt-Lite, a minimal pipeline for optimizing LLM agent skills by treating rollout traces as filesystem files and applying trajectory exploration, consensus mining, and independent validation; integrates as a one-line VSCode Copilot command and reports cross-benchmark improvements that let smaller models sometimes outperform larger ones.
AI Video Papers2026
Provides a reflexive agentic framework for long-horizon video understanding that replaces costly iterative reasoning with dual contextual states: a consolidated global multimodal script and parametric latent states for fast retrieval and response, improving speed and memory efficiency.
AI Agent Papers2026
Evaluates proactive, multimodal agents on 400 bilingual real‑world tasks across five capability axes (Skill Usage, Exploration, Long‑Context Reasoning, Multimodal Understanding, Cross‑Platform Coordination) using live Docker‑based, stepwise closed‑loop evaluation to separate base model skills from framework design.
AI Agent Papers2026
Acquires repository knowledge via a targeted QA loop before generating patches, decoupling knowledge acquisition from repair. A Questioner and Answerer produce evidence-grounded QA pairs that a Resolver uses to generate fixes; improves Pass@1 on SWE-bench Verified with modest overhead.
Builds structured knowledge graphs for retrieval-augmented generation via a multi-step GraphRAG pipeline that separates extraction from consolidation. Key features include typed two-stage extraction, DBSCAN-backed deduplication, LLM summarization, Leiden community detection, and a compact 7B extractor model (Meno-Lite-0.1).
AI Agent Papers2026
Converts completed on-policy trajectories into natural-language 'hindsight skills' and converts the skill-induced action probability shifts into a dense token-level on-policy distillation signal, jointly optimized with outcome-based RL to improve sample efficiency and long-horizon agent behavior.
A PyTorch-native training framework for agentic reinforcement learning research that keeps researcher-facing code compact and editable. Uses an asynchronous loop to train multimodal and mixture-of-experts policies while never training on tokens the agent didn't generate; matches Megatron-style stacks under a comparable protocol and ships recipes and containers on GitHub.
Selects a referred target from candidate bounding boxes, then decodes tracking waypoints for single-camera embodied visual tracking. Injects past selected-bbox geometry via sliding-window TVBI tokens and is co-trained on a Refer‑QA dataset; achieves SOTA on EVT‑Bench and demonstrates sim-to-real on legged and humanoid robots.
AI Model2026
A sparsely activated Mixture-of-Experts (MoE) causal language model with 16B total parameters and 2.8B active parameters per token, released with end-to-end checkpoints and training recipe; trained on AMD Instinct GPUs and licensed for research use.
AI Dataset2026
Provides a near-deduplicated, quality-filtered 15.9 TB training subset of GitHub source code grouped by repository, with inline UTF‑8 file contents and repo metadata for pre-training and analysis of code LLMs; cutoff Aug 7, 2025, ODC-By license.
Presents Skill Self-Play (Skill-SP), a co-evolutionary training loop where a proposer, solver, and dynamic skill controller generate, solve, and verify tasks conditioned on reusable skills — balancing verifiable execution with open-ended task diversity to boost LLM tool-use and reasoning.
Transforms open-ended LLM optimization into self-verifiable reinforcement learning by turning tasks into proxy environments that produce deterministic, rule-based rewards. Proposes RLSVR and SpyRL — an information-asymmetric self-play scheme where agents vote to identify a preassigned spy, yielding verifiable rewards without human annotation. Demonstrated on summarization, creative writing and mathematical reasoning.