Equationliu / Kangaroo
Implementation of Kangaroo: Lossless Self-Speculative Decoding via Double Early Exiting
☆39Updated 2 months ago
Related projects: ⓘ
- GEAR: An Efficient KV Cache Compression Recipefor Near-Lossless Generative Inference of LLM☆134Updated 2 months ago
- QAQ: Quality Adaptive Quantization for LLM KV Cache☆42Updated 5 months ago
- ☆164Updated 4 months ago
- ☆75Updated this week
- ☆37Updated 5 months ago
- Boosting 4-bit inference kernels with 2:4 Sparsity☆47Updated 2 weeks ago
- Code for Palu: Compressing KV-Cache with Low-Rank Projection☆39Updated this week
- ☆102Updated 3 months ago
- Official repository for LightSeq: Sequence Level Parallelism for Distributed Training of Long Context Transformers☆183Updated last month
- ☆32Updated 7 months ago
- Low-bit optimizers for PyTorch☆109Updated 11 months ago
- [ICML 2024 Oral] This project is the official implementation of our Accurate LoRA-Finetuning Quantization of LLMs via Information Retenti…☆55Updated 5 months ago
- Ouroboros: Speculative Decoding with Large Model Enhanced Drafting☆60Updated 6 months ago
- This repo contains the source code for: Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs☆28Updated last month
- KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache☆213Updated 3 weeks ago
- Spec-Bench: A Comprehensive Benchmark and Unified Evaluation Platform for Speculative Decoding (ACL 2024 Findings)☆156Updated 3 months ago
- The official implementation of the EMNLP 2023 paper LLM-FP4☆156Updated 9 months ago
- [ICML 2024] Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference☆161Updated 2 months ago
- Official PyTorch implementation of IntactKV: Improving Large Language Model Quantization by Keeping Pivot Tokens Intact☆25Updated 3 months ago
- ☆130Updated last year
- Activation-aware Singular Value Decomposition for Compressing Large Language Models☆43Updated 3 weeks ago
- Official Pytorch Implementation of "Outlier Weighed Layerwise Sparsity (OWL): A Missing Secret Sauce for Pruning LLMs to High Sparsity"☆49Updated 2 months ago
- Code repo for the paper "SpinQuant LLM quantization with learned rotations"☆79Updated this week
- Unofficial implementation for the paper "Mixture-of-Depths: Dynamically allocating compute in transformer-based language models"☆123Updated 2 months ago
- Triton implementation of Flash Attention2.0☆21Updated last year
- AFPQ code implementation☆15Updated 10 months ago
- [MLSys'24] Atom: Low-bit Quantization for Efficient and Accurate LLM Serving☆258Updated 2 months ago
- Breaking Throughput-Latency Trade-off for Long Sequences with Speculative Decoding☆55Updated this week
- [ACL 2024] RelayAttention for Efficient Large Language Model Serving with Long System Prompts☆29Updated 6 months ago
- Official implementation of the ICLR 2024 paper AffineQuant☆16Updated 5 months ago