deepseek-ai / DualPipeLinks
A bidirectional pipeline parallelism algorithm for computation-communication overlap in V3/R1 training.
☆2,851Updated 5 months ago
Alternatives and similar repositories for DualPipe
Users that are interested in DualPipe are comparing it to the libraries listed below
Sorting:
- DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling☆5,628Updated last week
- Expert Parallelism Load Balancer☆1,250Updated 4 months ago
- Analyze computation-communication overlap in V3/R1.☆1,090Updated 5 months ago
- DeepEP: an efficient expert-parallel communication library☆8,422Updated this week
- FlashMLA: Efficient MLA kernels☆11,699Updated last week
- Production-tested AI infrastructure tools for efficient AGI development and community-driven innovation☆7,901Updated 3 months ago
- Mooncake is the serving platform for Kimi, a leading LLM service provided by Moonshot AI.☆3,768Updated last week
- A high-performance distributed file system designed to address the challenges of AI training and inference workloads.☆9,238Updated this week
- FlashInfer: Kernel Library for LLM Serving☆3,571Updated this week
- MoBA: Mixture of Block Attention for Long-Context LLMs☆1,870Updated 4 months ago
- Muon is Scalable for LLM Training☆1,281Updated 2 weeks ago
- A lightweight data processing framework built on DuckDB and 3FS.☆4,764Updated 5 months ago
- A fast communication-overlapping library for tensor/expert parallelism on GPUs.☆1,072Updated this week
- DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models☆1,769Updated last year
- High-performance inference framework for large language models, focusing on efficiency, flexibility, and availability.☆1,243Updated this week
- Distributed RL System for LLM Reasoning☆2,393Updated this week
- The official repo of MiniMax-Text-01 and MiniMax-VL-01, large-language-model & vision-language-model based on Linear Attention☆3,123Updated last month
- A Datacenter Scale Distributed Inference Serving Framework☆4,786Updated this week
- Domain-specific language designed to streamline the development of high-performance GPU/CPU/Accelerators kernels☆1,530Updated this week
- Simple RL training for reasoning☆3,728Updated 2 weeks ago
- Mirage: Automatically Generating Fast GPU Kernels without Programming in Triton/CUDA☆1,688Updated this week
- 🐳 Efficient Triton implementations for "Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention"☆822Updated 5 months ago
- slime is a LLM post-training framework aiming for RL Scaling.☆1,375Updated this week
- Democratizing Reinforcement Learning for LLMs☆4,043Updated this week
- Distributed Compiler based on Triton for Parallel Systems☆1,031Updated last week
- Fast, Flexible and Portable Structured Generation☆1,172Updated last week
- ☆1,180Updated last month
- A PyTorch Native LLM Training Framework☆861Updated last month
- Official Repo for Open-Reasoner-Zero☆2,024Updated 2 months ago
- A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit floating point (FP8) precision on Hopper, Ada and Bla…☆2,645Updated last week