mit-han-lab / duo-attentionView external linksLinks
[ICLR 2025] DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads
☆524Feb 10, 2025Updated last year
Alternatives and similar repositories for duo-attention
Users that are interested in duo-attention are comparing it to the libraries listed below
Sorting:
- [ICML 2024] Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference☆372Jul 10, 2025Updated 7 months ago
- A sparse attention kernel supporting mix sparse patterns☆455Jan 18, 2026Updated 3 weeks ago
- [MLSys'25] QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving; [MLSys'25] LServe: Efficient Long-sequence LLM Se…☆812Mar 6, 2025Updated 11 months ago
- [NeurIPS'24 Spotlight, ICLR'25, ICML'25] To speed up Long-context LLMs' inference, approximate and dynamic sparse calculate the attention…☆1,183Sep 30, 2025Updated 4 months ago
- [ICML 2025] XAttention: Block Sparse Attention with Antidiagonal Scoring☆269Jul 6, 2025Updated 7 months ago
- ☆302Jul 10, 2025Updated 7 months ago
- Quantized Attention on GPU☆44Nov 22, 2024Updated last year
- [ICML 2024] KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache☆356Nov 20, 2025Updated 2 months ago
- 📰 Must-read papers on KV Cache Compression (constantly updating 🤗).☆658Sep 30, 2025Updated 4 months ago
- GEAR: An Efficient KV Cache Compression Recipefor Near-Lossless Generative Inference of LLM☆176Jul 12, 2024Updated last year
- The code of our paper "InfLLM: Unveiling the Intrinsic Capacity of LLMs for Understanding Extremely Long Sequences with Training-Free Mem…☆397Apr 20, 2024Updated last year
- [COLM 2024] TriForce: Lossless Acceleration of Long Sequence Generation with Hierarchical Speculative Decoding☆276Aug 31, 2024Updated last year
- [ICLR2025 Spotlight] MagicPIG: LSH Sampling for Efficient LLM Generation☆249Dec 16, 2024Updated last year
- [MLSys'24] Atom: Low-bit Quantization for Efficient and Accurate LLM Serving☆336Jul 2, 2024Updated last year
- The official implementation of paper: SimLayerKV: A Simple Framework for Layer-Level KV Cache Reduction.☆52Oct 18, 2024Updated last year
- open-source code for paper: Retrieval Head Mechanistically Explains Long-Context Factuality☆231Aug 2, 2024Updated last year
- [NeurIPS'23] H2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models.☆502Aug 1, 2024Updated last year
- [ICLR2025] Breaking Throughput-Latency Trade-off for Long Sequences with Speculative Decoding☆142Dec 4, 2024Updated last year
- [ICML 2025 Spotlight] ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inference☆283May 1, 2025Updated 9 months ago
- An algorithm for weight-activation quantization (W4A4, W4A8) of LLMs, supporting both static and dynamic quantization☆172Nov 26, 2025Updated 2 months ago
- FlashInfer: Kernel Library for LLM Serving☆4,935Updated this week
- FP16xINT4 LLM inference kernel that can achieve near-ideal ~4x speedups up to medium batchsizes of 16-32 tokens.☆1,011Sep 4, 2024Updated last year
- [ICLR 2025] TidalDecode: A Fast and Accurate LLM Decoding with Position Persistent Sparse Attention☆52Aug 6, 2025Updated 6 months ago
- A throughput-oriented high-performance serving framework for LLMs☆945Oct 29, 2025Updated 3 months ago
- 16-fold memory access reduction with nearly no loss☆110Mar 26, 2025Updated 10 months ago
- KV cache compression for high-throughput LLM inference☆154Feb 5, 2025Updated last year
- [NeurIPS 2024] KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization☆404Aug 13, 2024Updated last year
- Official Implementation of EAGLE-1 (ICML'24), EAGLE-2 (EMNLP'24), and EAGLE-3 (NeurIPS'25).☆2,180Jan 27, 2026Updated 2 weeks ago
- Code for paper: [ICLR2025 Oral] FlexPrefill: A Context-Aware Sparse Attention Mechanism for Efficient Long-Sequence Inference☆160Oct 13, 2025Updated 4 months ago
- Layer-Condensed KV cache w/ 10 times larger batch size, fewer params and less computation. Dramatic speed up with better task performance…☆157Apr 7, 2025Updated 10 months ago
- ☆65Apr 26, 2025Updated 9 months ago
- BitBLAS is a library to support mixed-precision matrix multiplications, especially for quantized LLM deployment.☆752Aug 6, 2025Updated 6 months ago
- Efficient LLM Inference over Long Sequences☆392Jun 25, 2025Updated 7 months ago
- 🐳 Efficient Triton implementations for "Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention"☆965Feb 5, 2026Updated last week
- [ICLR 2025] Palu: Compressing KV-Cache with Low-Rank Projection☆155Feb 20, 2025Updated 11 months ago
- The Official Implementation of Ada-KV [NeurIPS 2025]☆129Nov 26, 2025Updated 2 months ago
- [ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-t…☆3,159Jan 17, 2026Updated 3 weeks ago
- Model Compression Toolbox for Large Language Models and Diffusion Models☆753Aug 14, 2025Updated 6 months ago
- ☆352Apr 2, 2024Updated last year