Trillion-Parameter Multimodal, 4B Agents Match 671B, PPO Exposed

Featured

• ERNIE 5.0 Technical Report score 9
  入选 HF Daily Papers; HF 热度: 235 upvotes (+4); 关键词(11): scaling, efficient, deployment, production, inference
• WideSeek-R1: Exploring Width Scaling for Broad Information Seeking via Multi-Agent Reinforcement Learning score 9
  入选 HF Daily Papers; HF 热度: 85 upvotes (+4); 关键词(5): scaling, agent, tool use, reasoning, benchmark
• SocialVeil: Probing Social Intelligence of Language Agents under Communication Barriers score 8
  入选 HF Daily Papers; HF 热度: 16 upvotes (+3); 关键词(2): agents, evaluation
• From Data to Behavior: Predicting Unintended Model Behaviors Before Training score 8
  入选 HF Daily Papers; HF 热度: 13 upvotes (+3); 关键词(5): lightweight, fine-tuning, pre-training, evaluation, safety
• Training Data Efficiency in Multimodal Process Reward Models score 9
  入选 HF Daily Papers; HF 热度: 73 upvotes (+4); 关键词(4): efficiency, reasoning, multimodal, cost
• OmniSIFT: Modality-Asymmetric Token Compression for Efficient Omni-modal Large Language Models score 9
  入选 HF Daily Papers; HF 热度: 44 upvotes (+4); 关键词(6): efficient, compression, pruning, latency, multimodal
• EgoActor: Grounding Task Planning into Spatial-aware Egocentric Actions for Humanoid Robots via Visual-Language Models score 9
  入选 HF Daily Papers; HF 热度: 33 upvotes (+4); 关键词(5): inference, real-time, reasoning, planning, vision-language
• Rethinking the Trust Region in LLM Reinforcement Learning score 9
  入选 HF Daily Papers; HF 热度: 29 upvotes (+4); 关键词(5): efficient, efficiency, serving, fine-tuning, PPO
• Privileged Information Distillation for Language Models score 9
  入选 HF Daily Papers; HF 热度: 20 upvotes (+4); 关键词(6): distillation, inference, finetuning, agents, agentic
• Protein Autoregressive Modeling via Multiscale Structure Generation score 9
  机构: ByteDance; 入选 HF Daily Papers; HF 热度: 3 upvotes (+1); 关键词(4): scaling, fine-tuning, transformer, benchmark

Also Worth Noting

• Skin Tokens: A Learned Compact Representation for Unified Autoregressive Rigging score 4
  入选 HF Daily Papers; HF 热度: 4 upvotes (+1)
• EntRGi: Entropy Aware Reward Guidance for Diffusion Language Models score 4
  入选 HF Daily Papers; 关键词(1): diffusion
• LORE: Jointly Learning the Intrinsic Dimensionality and Relative Similarity Structure From Ordinal Data score 4
  关键词(1): efficient; 顶会接收: ICLR
• AFD-INSTRUCTION: A Comprehensive Antibody Instruction Dataset with Functional Annotations for LLM-Based Understanding and Design score 4
  关键词(1): alignment; 顶会接收: ICLR
• SkeletonGaussian: Editable 4D Generation through Gaussian Skeletonization score 3
  入选 HF Daily Papers
• Comparative Insights on Adversarial Machine Learning from Industry and Academia: A User-Study Approach score 3
  机构: Carnegie Mellon
• Temporal Pair Consistency for Variance-Reduced Flow Matching score 2
  关键词(4): efficiency, lightweight, diffusion, cost
• VideoBrain: Learning Adaptive Frame Sampling for Long Video Understanding score 2
  关键词(3): agent, agents, vision-language
• CoRe: Context-Robust Remasking for Diffusion Language Models score 2
  关键词(3): inference, diffusion, reasoning
• Rethinking Perplexity: Revealing the Impact of Input Length on Perplexity Evaluation in LLMs score 2
  关键词(6): efficiency, quantization, deployment, latency, evaluation
• Interfaze: The Future of AI is built on Task-Specific Small Models score 2
  关键词(2): transformer, multimodal
• DMS2F-HAD: A Dual-branch Mamba-based Spatial-Spectral Fusion Network for Hyperspectral Anomaly Detection score 2
  关键词(5): efficiency, inference, mamba, benchmark, cost
• Learning Where It Matters: Geometric Anchoring for Robust Preference Alignment score 2
  关键词(4): DPO, alignment, preference, reasoning
• Expert Selections In MoE Models Reveal (Almost) As Much As Text score 2
  关键词(3): inference, transformer, MoE