tsinghua-ideal / TwilightLinks
[NeurIPS'25 Spotlight] Adaptive Attention Sparsity with Hierarchical Top-p Pruning
☆67Updated last week
Alternatives and similar repositories for Twilight
Users that are interested in Twilight are comparing it to the libraries listed below
Sorting:
- 16-fold memory access reduction with nearly no loss☆107Updated 7 months ago
- ArkVale: Efficient Generative LLM Inference with Recallable Key-Value Eviction (NIPS'24)☆44Updated 11 months ago
- [ICLR 2025] TidalDecode: A Fast and Accurate LLM Decoding with Position Persistent Sparse Attention☆48Updated 3 months ago
- [COLM 2024] SKVQ: Sliding-window Key and Value Cache Quantization for Large Language Models☆24Updated last year
- [NeurIPS 2025] ClusterFusion: Expanding Operator Fusion Scope for LLM Inference via Cluster-Level Collective Primitive☆49Updated last month
- Code for paper: [ICLR2025 Oral] FlexPrefill: A Context-Aware Sparse Attention Mechanism for Efficient Long-Sequence Inference☆154Updated last month
- SpInfer: Leveraging Low-Level Sparsity for Efficient Large Language Model Inference on GPUs☆59Updated 7 months ago
- ☆143Updated this week
- ☆58Updated last year
- The Official Implementation of Ada-KV [NeurIPS 2025]☆113Updated last month
- [HPCA 2025] A GPU-optimized system for efficient long-context LLMs decoding with low-bit KV cache.☆62Updated last week
- [DAC'25] Official implement of "HybriMoE: Hybrid CPU-GPU Scheduling and Cache Management for Efficient MoE Inference"☆89Updated 5 months ago
- [NeurIPS 2024] Efficient LLM Scheduling by Learning to Rank☆64Updated last year
- ☆39Updated 3 months ago
- [ICML 2024] Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference☆353Updated 4 months ago
- Tile-Based Runtime for Ultra-Low-Latency LLM Inference☆178Updated this week
- [ICLR 2025] Palu: Compressing KV-Cache with Low-Rank Projection☆146Updated 9 months ago
- ☆27Updated 7 months ago
- Building the Virtuous Cycle for AI-driven LLM Systems☆91Updated this week
- ☆57Updated last year
- ☆79Updated last month
- Distributed MoE in a Single Kernel [NeurIPS '25]☆125Updated last month
- Scalable long-context LLM decoding that leverages sparsity—by treating the KV cache as a vector storage system.☆98Updated 2 months ago
- SeerAttention: Learning Intrinsic Sparse Attention in Your LLMs☆171Updated last month
- Implement some method of LLM KV Cache Sparsity☆42Updated last year
- Pytorch implementation of our paper accepted by ICML 2024 -- CaM: Cache Merging for Memory-efficient LLMs Inference☆47Updated last year
- [ACL 2025] Squeezed Attention: Accelerating Long Prompt LLM Inference☆54Updated last year
- ☆50Updated 6 months ago
- GEAR: An Efficient KV Cache Compression Recipefor Near-Lossless Generative Inference of LLM☆170Updated last year
- Explore Inter-layer Expert Affinity in MoE Model Inference☆15Updated last year