… The following papers were recommended by the Semantic Scholar API Reinforcing Consistency in Video MLLMs with Structured Rewards 2026 CycleCap: Improving VLMs Captioning Performance via Self-Supervised Cycle Consistency Fine-Tuning 2026 Building a Precise Video Language with Human-AI Oversight 2026… …
… We further propose Area Under the Rank Accuracy Curve AURAC , a metric that consistently evaluates the performance of hierarchical low-rank adapters. …
… Second, different one-shot expert compression methods converge to similar final performance after large-scale continual pretraining. Motivated by this, we introduce a simple partial-preservation expert merging strategy that improves downstream performance across most benchmarks. …
… We propose ROMA, an RL fine-tuning framework that modifies the optimization dynamics to reinforce reasoning against visual degradation while preserving clean-input performance. …
Papers arxiv:2605.09241 Sub-JEPA: Subspace Gaussian Regularization for Stable End-to-End World Models Published on May 10 Submitted by Kai Zhao on May 12 shanghai university Authors: Kai Zhao , , , , , , Abstract Joint-Embedding Predictive Architectures training is improved by applying Gaussian con… …
… Experiments across multiple model scales and architectures show that SEIF consistently improves instruction-following performance, suggesting strong generality. …
… Reaches competitive performance in just 80 epochs compared to 800+ epochs for previous methods. …
… Especially low-resource languages exhibit much lower reasoning performance. …
Papers arxiv:2605.00425 AEM: Adaptive Entropy Modulation for Multi-Turn Agentic Reinforcement Learning Published on May 8 Submitted by daxiang on May 11 BAIDU Authors: , , , , , , , , , , , Abstract A novel supervision-free credit assignment method for reinforcement learning in language model agent… …
… Token-level supervised fine-tuning SFT and point-wise regression rewards bias learning toward high-density regions, leading to regression-to-the-mean behavior and poor tail performance. …