🔴 High Significance
Model Releases
🔴 Maximal Brain Damage Without Data or Optimization: Disrupting Neural Networks via Sign-Bit Flips — score 80
Sources: huggingface
Deep Neural Networks (DNNs) can be catastrophically disrupted by flipping only a handful of parameter bits. We introduce Deep Neural Lesion (DNL), a data-free and optimizationfree method that locates critical parameters, and an enhanced single-pass variant, 1P-DNL, that refines this selection with o
Developer Tools
🔴 Elucidating the SNR-t Bias of Diffusion Probabilistic Models — score 95
Sources: huggingface
Diffusion Probabilistic Models have demonstrated remarkable performance across a wide range of generative tasks. However, we have observed that these models often suffer from a Signal-to-Noise Ratio-timestep (SNR-t) bias. This bias refers to the misalignment between the SNR of the denoising sample a
🔴 DiPO: Disentangled Perplexity Policy Optimization for Fine-grained Exploration-Exploitation Trade-Off — score 80
Sources: huggingface
Reinforcement Learning with Verifiable Rewards (RLVR) has catalyzed significant advances in the reasoning capabilities of Large Language Models (LLMs). However, effectively managing the exploration and exploitation trade-off remains a critical challenge. In this paper, we fully analyze the explorati
🟡 Notable
Model Releases
🟡 Qwen3.5-Omni Technical Report — score 65
Sources: huggingface
In this work, we present Qwen3.5-Omni, the latest advancement in the Qwen-Omni model family. Representing a significant evolution over its predecessor, Qwen3.5-Omni scales to hundreds of billions of parameters and supports a 256k context length. By leveraging a massive dataset comprising heterogeneo
🟡 PersonaVLM: Long-Term Personalized Multimodal LLMs — score 55
Sources: huggingface
Multimodal Large Language Models (MLLMs) serve as daily assistants for millions. However, their ability to generate responses aligned with individual preferences remains limited. Prior approaches enable only static, single-turn personalization through input augmentation or output alignment, and thus
🟡 OpenAI helps Hyatt advance AI among colleagues — score 50
Sources: lab_blog/OpenAI
Hyatt deploys ChatGPT Enterprise across its global workforce, using GPT-5.4 and Codex to improve productivity, operations, and guest experiences.
Developer Tools
🟡 Web Retrieval-Aware Chunking (W-RAC) for Efficient and Cost-Effective Retrieval-Augmented Generation Systems — score 45
Sources: huggingface
Retrieval-Augmented Generation (RAG) systems critically depend on effective document chunking strategies to balance retrieval quality, latency, and operational cost. Traditional chunking approaches, such as fixed-size, rule-based, or fully agentic chunking, often suffer from high token consumption,
🟢 Incremental
Model Releases
🟢 Cut Your Losses! Learning to Prune Paths Early for Efficient Parallel Reasoning — score 30
Sources: huggingface
Parallel reasoning enhances Large Reasoning Models (LRMs) but incurs prohibitive costs due to futile paths caused by early errors. To mitigate this, path pruning at the prefix level is essential, yet existing research remains fragmented without a standardized framework. In this work, we propose the
Developer Tools
🟢 Mind DeepResearch Technical Report — score 30
Sources: huggingface
We present Mind DeepResearch (MindDR), an efficient multi-agent deep research framework that achieves leading performance with only ~30B-parameter models through a meticulously designed data synthesis and multi-stage training pipeline. The core innovation of MindDR lies in a collaborative three-agen
🟢 Motif-Video 2B: Technical Report — score 5
Sources: huggingface
Training strong video generation models usually requires massive datasets, large parameter counts, and substantial compute. In this work, we ask whether strong text-to-video quality is possible at a much smaller budget: fewer than 10M clips and less than 100,000 H200 GPU hours. Our core claim is tha
Infrastructure & Compute
🟢 Where does output diversity collapse in post-training? — score 15
Sources: huggingface
Post-trained language models produce less varied outputs than their base counterparts. This output diversity collapse undermines inference-time scaling methods that rely on varied samples, and risks homogenizing model outputs on creative and value-laden tasks. Prior work attributes collapse to speci
📄 New Papers
| Title | Category | Score | Link |
|---|---|---|---|
| Elucidating the SNR-t Bias of Diffusion Probabilistic Models | developer_tool | 77 | Open |
| DiPO: Disentangled Perplexity Policy Optimization for Fine-grained Exploration-Exploitation Trade-Off | developer_tool | 64 | Open |
| Maximal Brain Damage Without Data or Optimization: Disrupting Neural Networks via Sign-Bit Flips | model_release | 64 | Open |
| Qwen3.5-Omni Technical Report | model_release | 59 | Open |
| PersonaVLM: Long-Term Personalized Multimodal LLMs | model_release | 48 | Open |
| Towards Intelligent Legal Document Analysis: CNN-Driven Classification of Case Law Texts | cs.AI | 0 | Open |
| Semantic Entanglement in Vector-Based Retrieval: A Formal Framework and Context-Conditioned Disentanglement Pipeline for Agentic RAG Systems | cs.AI | 0 | Open |
| SafeAnchor: Preventing Cumulative Safety Erosion in Continual Domain Adaptation of Large Language Models | cs.AI | 0 | Open |
| CAPO: Counterfactual Credit Assignment in Sequential Cooperative Teams | cs.AI | 0 | Open |
| Stratagem: Learning Transferable Reasoning via Trajectory-Modulated Game Self-Play | cs.AI | 0 | Open |
| WISV: Wireless-Informed Semantic Verification for Distributed Speculative Decoding in Device-Edge LLM Inference | cs.AI | 0 | Open |
| Before You Interpret the Profile: Validity Scaling for LLM Metacognitive Self-Report | cs.AI | 0 | Open |
| Co-evolving Agent Architectures and Interpretable Reasoning for Automated Optimization | cs.AI | 0 | Open |
| Screen Before You Interpret: A Portable Validity Protocol for Benchmark-Based LLM Confidence Signals | cs.AI | 0 | Open |
| Concurrent Criterion Validation of a Validity Screen for LLM Confidence Signals via Selective Prediction | cs.AI | 0 | Open |