AnswerDotAI / cold-compressLinks
Cold Compress is a hackable, lightweight, and open-source toolkit for creating and benchmarking cache compression methods built on top of GPT-Fast, a simple, PyTorch-native generation codebase.
☆137Updated 10 months ago
Alternatives and similar repositories for cold-compress
Users that are interested in cold-compress are comparing it to the libraries listed below
Sorting:
- ☆130Updated 4 months ago
- ☆198Updated 6 months ago
- Triton-based implementation of Sparse Mixture of Experts.☆219Updated 6 months ago
- Boosting 4-bit inference kernels with 2:4 Sparsity☆79Updated 9 months ago
- GEAR: An Efficient KV Cache Compression Recipefor Near-Lossless Generative Inference of LLM☆163Updated 11 months ago
- Simple implementation of Speculative Sampling in NumPy for GPT-2.☆95Updated last year
- ☆256Updated last year
- Explorations into some recent techniques surrounding speculative decoding☆269Updated 6 months ago
- Layer-Condensed KV cache w/ 10 times larger batch size, fewer params and less computation. Dramatic speed up with better task performance…☆149Updated 2 months ago
- ☆126Updated last year
- Repo for "LoLCATs: On Low-Rank Linearizing of Large Language Models"☆239Updated 4 months ago
- Unofficial implementation for the paper "Mixture-of-Depths: Dynamically allocating compute in transformer-based language models"☆163Updated last year
- [ICLR2025] Breaking Throughput-Latency Trade-off for Long Sequences with Speculative Decoding☆116Updated 6 months ago
- The official repo for "LLoCo: Learning Long Contexts Offline"☆118Updated last year
- Tree Attention: Topology-aware Decoding for Long-Context Attention on GPU clusters☆126Updated 6 months ago
- Official repository for DistFlashAttn: Distributed Memory-efficient Attention for Long-context LLMs Training☆210Updated 10 months ago
- KV cache compression for high-throughput LLM inference☆131Updated 4 months ago
- Implementation of Speculative Sampling as described in "Accelerating Large Language Model Decoding with Speculative Sampling" by Deepmind☆98Updated last year
- CUDA and Triton implementations of Flash Attention with SoftmaxN.☆70Updated last year
- Repository for Sparse Finetuning of LLMs via modified version of the MosaicML llmfoundry☆42Updated last year
- The official implementation of the paper "Towards Efficient Mixture of Experts: A Holistic Study of Compression Techniques (TMLR)".☆71Updated 3 months ago
- [ICML 2024] KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache☆303Updated 5 months ago
- The simplest implementation of recent Sparse Attention patterns for efficient LLM inference.☆70Updated last week
- REST: Retrieval-Based Speculative Decoding, NAACL 2024☆204Updated 6 months ago
- ☆114Updated 3 weeks ago
- Fast and memory-efficient exact attention☆68Updated 3 months ago
- Code for studying the super weight in LLM☆107Updated 6 months ago
- Official implementation for Yuan & Liu & Zhong et al., KV Cache Compression, But What Must We Give in Return? A Comprehensive Benchmark o…☆79Updated 3 months ago
- A minimal implementation of vllm.☆43Updated 10 months ago
- Flash-Muon: An Efficient Implementation of Muon Optimizer☆131Updated last week