… In most current agent systems, however, skills are still represented by text-heavy artifacts, including SKILL.md-style documents and structured records whose machine-usable evidence remains embedded largely in natural-language descriptions. …
… To address this issue, we propose a search-based memory framework called BOOKMARKS , which actively initializes, maintains, and updates task-relevant pieces of bookmarks for the current task e.g., character acting . …
… Motivated by the "search, select, update" principle underlying GDS, we propose Dynamic Latent Routing DLR , a language-model post-training method that jointly learns discrete latent codes , routing policies , and model parameters through dynamic search in a single training stage. …
… The following papers were recommended by the Semantic Scholar API Towards grounded autonomous research: an end-to-end LLM mini research loop on published computational physics 2026 PRBench: End-to-end Paper Reproduction in Physics Research 2026 ARIS: Autonomous Research via Adversarial Multi-Agent … …
… The following papers were recommended by the Semantic Scholar API PRAISE: Prefix-Based Rollout Reuse in Agentic Search Training 2026 Train at Moving Edge: Online-Verified Prompt Selection for Efficient RL Training of Large Reasoning Model 2026 Accelerating RL Post-Training Rollouts via System-Integ… …
… View arXiv page View PDF Add to collection Community TLDR: RubricEM introduces a rubric-guided reinforcement learning framework for training long-form deep research agents, enabling finer-grained stagewise credit assignment and reflection meta-policy training beyond verifiable rewards. noticed a po… …
… View arXiv page View PDF GitHub 7 Add to collection Community Closed-loop auto research turns agent-written code, real experiments, and evaluator feedback into an autonomous feedback loop that develops non-trivial training recipes. the most interesting nugget here is how lineage feedback lets evalu… …