… On diverse reasoning tasks and LLM backbones, LPO consistently improves training performance over typical policy gradient baselines under matched targets, while intrinsically preserving optimization stability and response diversity. …
… On established Multimodal Tabular Learning benchmarks, we show that tuning the embeddings to the task improves performance. …
… Experiments across open-source LLMs demonstrate performance gains up to 27% across evaluation modes over standard baselines. …
Papers arxiv:2604.26326 Addressing Performance Saturation for LLM RL via Precise Entropy Curve Control Published on May 10 Submitted by Bolian Li on May 12 Purdue University Authors: Bolian Li , Yifan Wang , , , , Abstract Entrocraft, a rejection-sampling approach for reinforcement learning, addres… …
…the target region and ground-truth mask are preserved, while attributes such as surface appearance, context, or material composition are modified to introduce misleading semantic cues. The benchmark contains 2,146 paired…
… Experiments show that Pixel-Searcher achieves the strongest open-source performance across all three task views, while failures mainly arise from evidence acquisition, identity resolution, and visual instance binding . …
…This framework leverages the internal geometric prior of the target modality and performs bounded correction on source-modality representations, thereby constructing substitute representations in the target modality. Experiments confirm its benefits in…
… At present, addressing these questions typically requires costly training runs whose aggregate performance metrics obscure the dynamics at the level of individual tokens. …
… Although frontier models achieve relatively strong performance on direct expression evidence, performance drops substantially on functional, indirect, and weak-support evidence, with evidence-type confusion emerging as a dominant failure mode. …
… Experiments on AppWorld, BFCL v3, and MCP-Universe show that Preping substantially improves over a no-memory baseline and achieves performance competitive with strong playbook-based methods built from offline or online experience, with deployment cost 2.99times lower on AppWorld and 2.23times lower… …