NVIDIA / kvpress
LLM KV cache compression made easy
☆463Updated last week
Alternatives and similar repositories for kvpress:
Users that are interested in kvpress are comparing it to the libraries listed below
- Efficient LLM Inference over Long Sequences☆370Updated this week
- [MLSys'25] QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving; [MLSys'25] LServe: Efficient Long-sequence LLM Se…☆643Updated last month
- 📰 Must-read papers on KV Cache Compression (constantly updating 🤗).☆388Updated 2 weeks ago
- Applied AI experiments and examples for PyTorch☆262Updated last month
- Fast low-bit matmul kernels in Triton☆291Updated this week
- Perplexity GPU Kernels☆251Updated this week
- [NeurIPS 2024] KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization☆340Updated 8 months ago
- Dynamic Memory Management for Serving LLMs without PagedAttention☆354Updated last week
- [MLSys'24] Atom: Low-bit Quantization for Efficient and Accurate LLM Serving☆305Updated 9 months ago
- [ICML 2024] KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache☆288Updated 3 months ago
- 🚀 Efficiently (pre)training foundation models with native PyTorch features, including FSDP for training and SDPA implementation of Flash…☆240Updated last week
- A subset of PyTorch's neural network modules, written in Python using OpenAI's Triton.☆534Updated this week
- KernelBench: Can LLMs Write GPU Kernels? - Benchmark with Torch -> CUDA problems☆274Updated last week
- Code for "LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding", ACL 2024☆286Updated last week
- A throughput-oriented high-performance serving framework for LLMs☆797Updated 7 months ago
- [ICLR 2025] DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads☆453Updated 2 months ago
- 🚀 Collection of components for development, training, tuning, and inference of foundation models leveraging PyTorch native components.☆192Updated this week
- [ICML 2024] Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference☆273Updated 5 months ago
- A low-latency & high-throughput serving engine for LLMs☆346Updated last week
- Collection of kernels written in Triton language☆120Updated 3 weeks ago
- This repository contains the experimental PyTorch native float8 training UX☆223Updated 8 months ago
- ☆237Updated 11 months ago
- FP16xINT4 LLM inference kernel that can achieve near-ideal ~4x speedups up to medium batchsizes of 16-32 tokens.☆809Updated 7 months ago
- Cataloging released Triton kernels.☆217Updated 3 months ago
- Ring attention implementation with flash attention☆743Updated 2 weeks ago
- Zero Bubble Pipeline Parallelism☆383Updated 2 weeks ago
- ☆313Updated last year
- Explorations into some recent techniques surrounding speculative decoding☆259Updated 4 months ago
- Cold Compress is a hackable, lightweight, and open-source toolkit for creating and benchmarking cache compression methods built on top of…☆128Updated 8 months ago
- Implementation of 💍 Ring Attention, from Liu et al. at Berkeley AI, in Pytorch☆511Updated 6 months ago