SpeakerMem三项基准二元准确率47.9%—69.2%

重点关注

• The Tasteful Agent: Measuring and Improving Taste in Long-Horizon Tasks score 10
  入选 HF Daily Papers；HF 热度: 132 upvotes (+4)；有代码实现；关键词(1): reasoning
• Ovis-Embedding: Pushing the Frontiers of Universal Omni-Modal Embeddings score 10
  入选 HF Daily Papers；HF 热度: 45 upvotes (+4)；有代码实现；关键词(1): distillation
• Flash-dLLM: IO-Aware KV Caching and Parallel Decoding for Fast, Memory-Efficient Diffusion LLMs score 10
  入选 HF Daily Papers；HF 热度: 22 upvotes (+4)；有代码实现；关键词(2): deployment, reasoning
• RULER: Instance-aware Rubric Rewards for SVG Generation score 8
  入选 HF Daily Papers；HF 热度: 78 upvotes (+4)；关键词(1): vision-language
• StableVQ: Practical Guidelines for Stable Vector-Quantized Tokenizer Training score 8
  入选 HF Daily Papers；HF 热度: 30 upvotes (+4)；关键词(2): lightweight, quantization
• Lean Pool: An AI-Maintained Archive of Formalized Mathematics score 8
  入选 HF Daily Papers；HF 热度: 17 upvotes (+3)；有代码实现
• JEV-as-a-Judge: Accept When Confident, Escalate When Unsure score 7
  入选 HF Daily Papers；HF 热度: 27 upvotes (+4)
• SpeakerMem-R1: Speaker-Centered Dual-Track Memory for Multi-Party Dialogue score 10
  入选 HF Daily Papers；HF 热度: 82 upvotes (+4)；有代码实现；关键词(3): deployment, GRPO, leaderboard
• ImIR: Image-Instruction Tuning for All-in-One Image Restoration score 8
  入选 HF Daily Papers；HF 热度: 6 upvotes (+2)；有代码实现；关键词(4): scaling, lightweight, instruction tuning, vision-language
• Agensh: Scaling Organizational Intelligence to 1,024 Agents score 7
  入选 HF Daily Papers；HF 热度: 14 upvotes (+3)；关键词(3): scaling, latency, agentic

也值得关注

• LatentPort: Beyond KV Cache - Cross-Model Transfer of Recurrent Memory in Hybrid Language Models: A 4B-to-9B Hybrid-State Handoff Without Target Prefix Replay score 7
  入选 HF Daily Papers；HF 热度: 7 upvotes (+2)；有代码实现
• Entropy Can Flow, or It Can Guide. Be Entropy. LEDFlow: Introducing Entropy-guided Generation Order into Uniform Discrete Flow score 4
  机构: Cambridge；关键词(2): reasoning, text-to-image
• Matryoshka attribution: Learning to attribute language model outputs to representations and weights score 4
  机构: Stanford；关键词(2): finetuning, leaderboard
• Domain-Adaptive Pretraining Enhances Water Treatment Semantic Representation for Large-Scale Structured Literature Mining score 4
  机构: Microsoft；关键词(1): pretraining
• Optimizing Denoising Trajectories in dLLMs: A Lightweight Evolutionary Heuristic Approach score 4
  机构: Huawei；关键词(2): lightweight, reasoning
• Partially Observed Sparse Graphs: The Unknown Sampling Rate is a Tail Index score 4
  机构: Tsinghua；关键词(1): edge
• Peerify: Benchmarking Peer-Review Claim Verification score 3
  机构: University of Toronto
• Continuous Optimization for p-adic Models score 3
  机构: Amazon
• FISSION: Label Augmentation for Bot Detection score 3
  机构: Yale
• Receptiveness, Not Sycophancy: Distinguishing Engagement from Deference in Language Models score 3
  机构: Stanford
• FleXray: Universal Clinical X-ray Segmentation score 3
  机构: MIT