DeepSpec: a full-stack codebase for training and evaluating speculative decoding algorithms
☆7,224Jul 9, 2026Updated 3 months ago
Alternatives and similar repositories for DeepSpec
Users that are interested in DeepSpec are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- DFlash: Block Diffusion for Flash Speculative Decoding☆6,178Aug 18, 2026Updated 2 months ago
- Train speculative decoding models effortlessly and port them smoothly to SGLang serving.☆1,225Updated this week
- slime is an LLM training framework for RL Scaling.☆8,682Updated this week
- FlashKDA: high-performance Kimi Delta Attention kernels☆1,281Sep 1, 2026Updated last month
- A kernel library written in tilelang☆1,962Updated this week
- Virtual machines for every use case on DigitalOcean • AdGet dependable uptime with 99.99% SLA, simple security tools, and predictable monthly pricing with DigitalOcean's virtual machines, called Droplets.
- TokenSpeed is a speed-of-light LLM inference engine.☆2,699Updated this week
- [EMNLP2026 Main] Official implementation of “Domino: Decoupling Causal Modeling from Autoregressive Drafting in Speculative Decoding”.☆139Jul 25, 2026Updated 2 months ago
- Domain-specific language designed to streamline the development of high-performance GPU/CPU/Accelerators kernels☆8,647Updated this week
- SGLang is a high-performance serving framework for large language models and multimodal models.☆37,274Updated this week
- DeepGEMM: clean and efficient BLAS kernel library on GPU☆8,980Updated this week
- FlashInfer: Kernel Library for LLM Serving☆6,672Updated this week
- JetSpec: Breaking the Scaling Ceiling of Speculative Decoding with Causal Parallel Tree Drafting☆238Updated this week
- Mooncake is the serving platform for Kimi, a leading LLM service provided by Moonshot AI.☆6,821Updated this week
- Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models☆4,761Jan 14, 2026Updated 9 months ago
- Proton VPN Special Offer - Get 70% off • AdSpecial partner offer. Trusted by over 100 million users worldwide. Tested, Approved and Recommended by Experts.
- high-performance linear attention kernel library built on TileLang☆728Updated this week
- Tile-Based Runtime for Ultra-Low-Latency LLM Inference☆1,866Sep 24, 2026Updated 3 weeks ago
- LMCache: Supercharge Your LLM with the Fastest KV Cache Layer☆12,108Updated this week
- A unified library for building, evaluating, and storing speculative decoding algorithms for LLM inference in vLLM☆888Updated this week
- Miles is an enterprise-facing reinforcement learning framework for LLM and VLM post-training, forked from and co-evolving with slime.☆3,159Updated this week
- A high-throughput and memory-efficient inference and serving engine for LLMs☆94,061Updated this week
- Accelerating MoE with IO and Tile-aware Optimizations☆780Aug 29, 2026Updated last month
- Nano vLLM☆15,956Apr 26, 2026Updated 5 months ago
- ☆432Updated this week
- GPU virtual machines on DigitalOcean Gradient AI • AdGet to production fast with high-performance AMD and NVIDIA GPUs you can spin up in seconds. The definition of operational simplicity.
- MoonEP: A Perfectly Balanced Expert Parallelism Library via Dynamic Redundant Experts☆1,182Updated this week
- 🚀 Efficient implementations for emerging model architectures☆5,942Updated this week
- A compact implementation of SGLang, designed to demystify the complexities of modern LLM serving systems.☆5,339May 17, 2026Updated 5 months ago
- verl/HybridFlow: A Flexible and Efficient RL Post-Training Framework☆23,915Updated this week
- AgentENV (AENV) is a distributed platform for running agent environments at scale.☆3,646Updated this week
- Official Implementation of EAGLE-1 (ICML'24), EAGLE-2 (EMNLP'24), and EAGLE-3 (NeurIPS'25).☆2,555Feb 20, 2026Updated 7 months ago
- Distributed Compiler and Optimized Parallel Kernels☆1,569Sep 18, 2026Updated last month
- High Performance LLM Inference Operator Library☆1,183Sep 7, 2026Updated last month
- A framework for efficient model inference with omni-modality models☆7,234Updated this week
- 1-Click AI Models by DigitalOcean Gradient • AdDeploy popular AI models on DigitalOcean Gradient GPU virtual machines with just a single click. Zero configuration with optimized deployments.
- Kernel Design Agents (KDA) is a agent-centric workflow to write high-performance CUDA Kernels.☆1,407Oct 1, 2026Updated 2 weeks ago
- A PyTorch native library for training speculative decoding models☆262Sep 29, 2026Updated 2 weeks ago
- FlashMLA: Efficient Multi-head Latent Attention Kernels☆13,084Updated this week
- A scalable, agentic-first, and HuggingFace-native RL framework for research (9k lines).☆1,199Updated this week
- DeepEP: an efficient expert-parallel communication library☆10,289Updated this week
- Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦☆42,703Updated this week
- A Quirky Assortment of CuTe Kernels☆1,176Updated this week