🔴 High Significance
Developer Tools
🔴 CAR-bench: Evaluating the Consistency and Limit-Awareness of LLM Agents under Real-World Uncertainty — score 95
Sources: huggingface
Existing benchmarks for Large Language Model (LLM) agents focus on task completion under idealistic settings but overlook reliability in real-world, user-facing applications. In domains, such as in-car voice assistants, users often issue incomplete or ambiguous requests, creating intrinsic uncertain
🔴 DFlash: Block Diffusion for Flash Speculative Decoding — score 85
Sources: huggingface
Autoregressive large language models (LLMs) deliver strong performance but require inherently sequential decoding, leading to high inference latency and poor GPU utilization. Speculative decoding mitigates this bottleneck by using a fast draft model whose outputs are verified in parallel by the targ
🔴 Spider-Sense: Intrinsic Risk Sensing for Efficient Agent Defense with Hierarchical Adaptive Screening — score 75
Sources: huggingface
As large language models (LLMs) evolve into autonomous agents, their real-world applicability has expanded significantly, accompanied by new security challenges. Most existing agent defense mechanisms adopt a mandatory checking paradigm, in which security validation is forcibly triggered at predefin
🟡 Notable
Model Releases
🟡 MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents — score 65
Sources: huggingface
Most Large Language Model (LLM) agent memory systems rely on a small set of static, hand-designed operations for extracting memory. These fixed procedures hard-code human priors about what to store and how to revise memory, making them rigid under diverse interaction patterns and inefficient on long
Developer Tools
🟡 Length-Unbiased Sequence Policy Optimization: Revealing and Controlling Response Length Variation in RLVR — score 55
Sources: huggingface
Recent applications of Reinforcement Learning with Verifiable Rewards (RLVR) to Large Language Models (LLMs) and Vision-Language Models (VLMs) have demonstrated significant success in enhancing reasoning capabilities for complex tasks. During RLVR training, an increase in response length is often re
🟡 Context Forcing: Consistent Autoregressive Video Generation with Long Context — score 45
Sources: huggingface
Recent approaches to real-time long video generation typically employ streaming tuning strategies, attempting to train a long-context student using a short-context (memoryless) teacher. In these frameworks, the student performs long rollouts but receives supervision from a teacher limited to short 5
Other Signals
🟡 Making AI work for everyone, everywhere: our approach to localization — score 50
Sources: lab_blog/OpenAI
OpenAI shares its approach to AI localization, showing how globally shared frontier models can be adapted to local languages, laws, and cultures without compromising safety.
🟢 Incremental
Model Releases
🟢 Dr. Kernel: Reinforcement Learning Done Right for Triton Kernel Generations — score 25
Sources: huggingface
High-quality kernel is critical for scalable AI systems, and enabling LLMs to generate such code would advance AI development. However, training LLMs for this task requires sufficient data, a robust environment, and the process is often vulnerable to reward hacking and lazy optimization. In these ca
Developer Tools
🟢 Reinforced Attention Learning — score 35
Sources: huggingface
Post-training with Reinforcement Learning (RL) has substantially improved reasoning in Large Language Models (LLMs) via test-time scaling. However, extending this paradigm to Multimodal LLMs (MLLMs) through verbose rationales yields limited gains for perception and can even degrade performance. We
🟢 RISE-Video: Can Video Generators Decode Implicit World Rules? — score 10
Sources: huggingface
While generative video models have achieved remarkable visual fidelity, their capacity to internalize and reason over implicit world rules remains a critical yet under-explored frontier. To bridge this gap, we present RISE-Video, a pioneering reasoning-oriented benchmark for Text-Image-to-Video (TI2
🟢 Reinforcement World Model Learning for LLM-based Agents — score 10
Sources: huggingface
Large language models (LLMs) have achieved strong performance in language-centric tasks. However, in agentic settings, LLMs often struggle to anticipate action consequences and adapt to environment dynamics, highlighting the need for world-modeling capabilities in LLM-based agents. We propose Reinfo
📄 New Papers
| Title | Category | Score | Link |
|---|---|---|---|
| CAR-bench: Evaluating the Consistency and Limit-Awareness of LLM Agents under Real-World Uncertainty | developer_tool | 91 | Open |
| DFlash: Block Diffusion for Flash Speculative Decoding | developer_tool | 75 | Open |
| Spider-Sense: Intrinsic Risk Sensing for Efficient Agent Defense with Hierarchical Adaptive Screening | developer_tool | 73 | Open |
| MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents | model_release | 67 | Open |
| Length-Unbiased Sequence Policy Optimization: Revealing and Controlling Response Length Variation in RLVR | developer_tool | 57 | Open |
| One Bias After Another: Mechanistic Reward Shaping and Persistent Biases in Language Reward Models | cs.AI | 0 | Open |
| Pro-ZD: A Transferable Graph Neural Network Approach for Proactive Zero-Day Threats Mitigation | cs.AI | 0 | Open |
| Do LLMs Act Like Rational Agents? Measuring Belief Coherence in Probabilistic Decision Making | cs.AI | 0 | Open |
| Toward generative machine learning for boosting ensembles of climate simulations | cs.AI | 0 | Open |
| Robust Pre-Training of Medical Vision-and-Language Models with Domain-Invariant Multi-Modal Masked Reconstruction | cs.AI | 0 | Open |
| LatentChem: From Textual CoT to Latent Thinking in Chemical Reasoning | cs.AI | 0 | Open |
| Accelerating Vision Transformers on Brain Processing Unit | cs.AI | 0 | Open |
| Attention's Gravitational Field:A Power-Law Interpretation of Positional Correlation | cs.AI | 0 | Open |
| CALM: Class-Conditional Sparse Attention Vectors for Large Audio-Language Models | cs.AI | 0 | Open |
| The Condensate Theorem: Transformers are O(n), Not $O(n^2)$ | cs.AI | 0 | Open |