whyNLP / LCKV
Layer-Condensed KV cache w/ 10 times larger batch size, fewer params and less computation. Dramatic speed up with better task performance. Accepted to ACL 2024.
☆139Updated this week
Related projects ⓘ
Alternatives and complementary repositories for LCKV
- Breaking Throughput-Latency Trade-off for Long Sequences with Speculative Decoding☆79Updated this week
- ☆122Updated 10 months ago
- Tree Attention: Topology-aware Decoding for Long-Context Attention on GPU clusters☆104Updated 2 months ago
- Cold Compress is a hackable, lightweight, and open-source toolkit for creating and benchmarking cache compression methods built on top of…☆87Updated 3 months ago
- Code for exploring Based models from "Simple linear attention language models balance the recall-throughput tradeoff"☆214Updated this week
- ☆189Updated 6 months ago
- Experiments on speculative sampling with Llama models☆118Updated last year
- Positional Skip-wise Training for Efficient Context Window Extension of LLMs to Extremely Length (ICLR 2024)☆199Updated 6 months ago
- Ouroboros: Speculative Decoding with Large Model Enhanced Drafting (EMNLP 2024 main)☆77Updated last month
- Unofficial implementation for the paper "Mixture-of-Depths: Dynamically allocating compute in transformer-based language models"☆135Updated 5 months ago
- ☆96Updated 2 months ago
- The source code of our work "Prepacking: A Simple Method for Fast Prefilling and Increased Throughput in Large Language Models"☆56Updated last month
- This is the official repository for Inheritune.☆105Updated last month
- Simple implementation of Speculative Sampling in NumPy for GPT-2.☆89Updated last year
- ☆64Updated last month
- Official PyTorch implementation of DistiLLM: Towards Streamlined Distillation for Large Language Models (ICML 2024)☆139Updated 2 months ago
- The official repo for "LLoCo: Learning Long Contexts Offline"☆113Updated 5 months ago
- Easy control for Key-Value Constrained Generative LLM Inference(https://arxiv.org/abs/2402.06262)☆58Updated 9 months ago
- GEAR: An Efficient KV Cache Compression Recipefor Near-Lossless Generative Inference of LLM☆149Updated 4 months ago
- Block Transformer: Global-to-Local Language Modeling for Fast Inference (Official Code)☆135Updated last month
- KV cache compression for high-throughput LLM inference☆87Updated last week
- Multipack distributed sampler for fast padding-free training of LLMs☆178Updated 3 months ago
- Pytorch implementation for "Compressed Context Memory For Online Language Model Interaction" (ICLR'24)☆50Updated 7 months ago
- Token Omission Via Attention☆121Updated last month
- open-source code for paper: Retrieval Head Mechanistically Explains Long-Context Factuality☆160Updated 3 months ago
- Code and Data for "Long-context LLMs Struggle with Long In-context Learning"☆91Updated 4 months ago
- [ICML 2024] KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache☆243Updated last month
- Simple and efficient pytorch-native transformer training and inference (batched)☆61Updated 7 months ago
- Official implementation for 'Extending LLMs’ Context Window with 100 Samples'☆74Updated 10 months ago
- Explorations into some recent techniques surrounding speculative decoding☆212Updated last year