…Multi-entity & Evolving Memory Evaluation Published on May 12 Submitted by Arnas Uselis on May 13 KAIST AI Authors: , , Arnas Uselis , , Abstract MEME benchmark evaluates memory systems across multiple entities and evolving…
…dynamic character development that evolves through narratives, necessitating benchmarks that evaluate psychological trajectory alignment rather than static factual recall, with ArcANE demonstrating superior performance when character arc information is conditioned into models…
…manually curated evaluation benchmark for rigorous assessment. We establish a multi-stage quality control process to ensure scientific validity and alignment across modalities. Experimental validation demonstrates significant performance improvements for models trained…
…Generated by Qwen/Qwen2.5-Coder-32B-Instruct World-action models (WAMs) jointly generate future video and robot actions through iterative diffusion, achieving strong performance on manipulation benchmarks but requiring tens of…
…Towards Improving Long-Context Performance with KV Cache Eviction Published on May 10 Submitted by Ngoc Bui on May 12 Yale University Authors: , , , Abstract Learned global retention-based key-value cache eviction…
…TT-OPD achieves the best performance on 10 of 18 benchmarks with an average +3.9~pp improvement over the non-RL baseline with faster early convergence, controlled response length, and sustained…
…However, accurately interpreting unconstrained STEM student handwritten solutions with intertwined mathematical formulas, diagrams, and textual reasoning poses a significant challenge due to the lack of authentic and domain-specific benchmarks. Additionally, current…
…An Intrinsic Capability of Attention-Based Models Published on May 8 Submitted by Elad Hoffer on May 14 NVIDIA Authors: , , , , , Abstract INTRA demonstrates that attention-based models can perform retrieval directly from…
…For parallelism, we benchmark four embedding models on FLORES-200 and BOUQuET retrieval tasks, covering 6,654 source--target directions in our target language-pair inventory. For QE, we evaluate nine reference…
…Sanitization confines the on-device LLM to proposing which spans to mask, while a deterministic registry performs all substitution and reversal, keeping actions directly executable on device. On three agentic benchmarks under…