YOLO-Master: MOE-Accelerated with Specialized Transformers for Enhanced Real-time Detection
Paper • 2512.23273 • Published • 15
papers to read
-
-
A 58-Addition, Rank-23 Scheme for General 3x3 Matrix Multiplication
Paper • 2512.21980 • Published • 3 -
Step-DeepResearch Technical Report
Paper • 2512.20491 • Published • 89 -
SAM Audio: Segment Anything in Audio
Paper • 2512.18099 • Published • 25 -
LoPA: Scaling dLLM Inference via Lookahead Parallel Decoding
Paper • 2512.16229 • Published • 17 -
CASA: Cross-Attention via Self-Attention for Efficient Vision-Language Fusion
Paper • 2512.19535 • Published • 13 -
Physics of Language Models: Part 4.1, Architecture Design and the Magic of Canon Layers
Paper • 2512.17351 • Published • 29 -
Fast and Accurate Causal Parallel Decoding using Jacobi Forcing
Paper • 2512.14681 • Published • 44 -
Janus: Disaggregating Attention and Experts for Scalable MoE Inference
Paper • 2512.13525 • Published • 6 -
QwenLong-L1.5: Post-Training Recipe for Long-Context Reasoning and Memory Management
Paper • 2512.12967 • Published • 113 -
Error-Free Linear Attention is a Free Lunch: Exact Solution from Continuous-Time Dynamics
Paper • 2512.12602 • Published • 44 -
Sliding Window Attention Adaptation
Paper • 2512.10411 • Published • 21 -
mHC: Manifold-Constrained Hyper-Connections
Paper • 2512.24880 • Published • 336
FlashSampling: Fast and Memory-Efficient Exact Sampling
Paper • 2603.15854 • Published • 9Note done
-
Attention Residuals
Paper • 2603.15031 • Published • 194 -
Mixture-of-Depths Attention
Paper • 2603.15619 • Published • 83
Flash-KMeans: Fast and Memory-Efficient Exact K-Means
Paper • 2603.09229 • Published • 85Note done
-
TriAttention: Efficient Long Reasoning with Trigonometric KV Compression
Paper • 2604.04921 • Published • 117 -
Flux Attention: Context-Aware Hybrid Attention for Efficient LLMs Inference
Paper • 2604.07394 • Published • 16 -
Sessa: Selective State Space Attention
Paper • 2604.18580 • Published • 14 -
Channel-wise Vector Quantization
Paper • 2605.26089 • Published • 15 -
LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decoding
Paper • 2605.27365 • Published • 147 -
MiniMax Sparse Attention
Paper • 2606.13392 • Published • 168 -
Thinking with Visual Grounding
Paper • 2606.16122 • Published • 11 -
KronQ: LLM Quantization via Kronecker-Factored Hessian
Paper • 2607.07964 • Published • 33 -
Jet-Long: Efficient Long-Context Extension with Dynamic Bifocal RoPE
Paper • 2607.07740 • Published • 26 -
Linear Attention Architectures: Mechanisms, Trade-offs, and Cross-Layer Routing
Paper • 2607.07953 • Published • 16 -
VideoChat3: Fully Open Video MLLM for Efficient and Generalist Video Understanding
Paper • 2607.14935 • Published • 173 -
OasisKV: Scaling In-Decode KV Cache Beyond HBM with Lookahead Sparse Prefetching
Paper • 2608.08097 • Published • 26 -
FlashPrefill V2: Block-Sparse Prefill Attention for Long-Context LLM Serving
Paper • 2608.19758 • Published • 20 -
Llama-Mobile: Efficient 2.7-Bit Quantization of VLMs
Paper • 2608.21134 • Published • 8 -
MaxKernel: Agentic Kernel Generation for TPUs
Paper • 2609.04523 • Published • 17 -
VC-Attention: Value Smoothing and Softmax Casting for Low-bit Attention
Paper • 2609.15810 • Published • 34