Best practices for training DeepSeek, Mixtral, Qwen and other MoE models using Megatron Core.
☆192May 20, 2026Updated this week
Alternatives and similar repositories for Megatron-MoE-ModelZoo
Users that are interested in Megatron-MoE-ModelZoo are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- Pipeline Parallelism Emulation and Visualization☆83Jan 8, 2026Updated 4 months ago
- ☆47Sep 8, 2025Updated 8 months ago
- Ongoing research training transformer models at scale☆18May 19, 2026Updated last week
- Accelerate LLM preference tuning via prefix sharing with a single line of code☆52Jul 4, 2025Updated 10 months ago
- NVIDIA Resiliency Extension is a python package for framework developers and users to implement fault-tolerant features. It improves the …☆291Updated this week
- Deploy on Railway without the complexity - Free Credits Offer • AdConnect your repo and Railway handles the rest with instant previews. Quickly provision container image services, databases, and storage volumes.
- Bridge Megatron-Core to Hugging Face/Reinforcement Learning☆214Updated this week
- Megatron's multi-modal data loader☆349Updated this week
- The official repo of Pai-Megatron-Patch for LLM & VLM large scale training developed by Alibaba Cloud.☆1,576Dec 15, 2025Updated 5 months ago
- PyTorch bindings for CUTLASS grouped GEMM.☆188Apr 8, 2026Updated last month
- Allow torch tensor memory to be released and resumed later☆245May 16, 2026Updated last week
- DeepSeek-V3.2-Exp DSA Warmup Lightning Indexer training operator based on tilelang☆44Nov 19, 2025Updated 6 months ago
- LLM training technologies developed by kwai☆72Jan 21, 2026Updated 4 months ago
- Cute layout visualization☆39Jan 18, 2026Updated 4 months ago
- Perplexity GPU Kernels☆576Nov 7, 2025Updated 6 months ago
- GPUs on demand by Runpod - Special Offer Available • AdRun AI, ML, and HPC workloads on powerful cloud GPUs—without limits or wasted spend. Deploy GPUs in under a minute and pay by the second.
- ☆32Jul 2, 2025Updated 10 months ago
- [WIP] Better (FP8) attention for Hopper☆33Feb 24, 2025Updated last year
- [Archived] For the latest updates and community contribution, please visit: https://github.com/Ascend/TransferQueue or https://gitcode.co…☆15Jan 16, 2026Updated 4 months ago
- A fast communication-overlapping library for tensor/expert parallelism on GPUs.☆1,309Aug 28, 2025Updated 8 months ago
- A Triton-only attention backend for vLLM☆25Mar 17, 2026Updated 2 months ago
- Ring attention implementation with flash attention☆1,020Sep 10, 2025Updated 8 months ago
- Sequence-level 1F1B schedule for LLMs.☆37Aug 26, 2025Updated 9 months ago
- ☆189May 15, 2026Updated last week
- NVSHMEM‑Tutorial: Build a DeepEP‑like GPU Buffer☆190Feb 11, 2026Updated 3 months ago
- Managed hosting for WordPress and PHP on Cloudways • AdManaged hosting for WordPress, Magento, Laravel, or PHP apps, on multiple cloud providers. Deploy in minutes on Cloudways by DigitalOcean.
- A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit and 4-bit floating point (FP8 and FP4) precision on H…☆3,350Updated this week
- ☆52May 19, 2025Updated last year
- Utility scripts for PyTorch (e.g. Make Perfetto show some disappearing kernels, Memory profiler that understands more low-level allocatio…☆107Sep 11, 2025Updated 8 months ago
- Zero Bubble Pipeline Parallelism☆456May 7, 2025Updated last year
- Distributed Compiler based on Triton for Parallel Systems☆1,440Apr 22, 2026Updated last month
- DeepXTrace is a lightweight tool for precisely diagnosing slow ranks in DeepEP-based environments.☆97Jan 16, 2026Updated 4 months ago
- ☆66Apr 26, 2025Updated last year
- Triton-based Symmetric Memory operators and examples☆100May 15, 2026Updated last week
- ☆365Jan 28, 2026Updated 3 months ago
- Managed Kubernetes at scale on DigitalOcean • AdDigitalOcean Kubernetes includes the control plane, bandwidth allowance, container registry, automatic updates, and more for free.
- Training library for Megatron-based models with bidirectional Hugging Face conversion capability☆642Updated this week
- Estimate MFU for DeepSeekV3☆26Jan 5, 2025Updated last year
- Accelerating MoE with IO and Tile-aware Optimizations☆691May 14, 2026Updated last week
- Pytorch implementation of our paper accepted by ICML 2024 -- CaM: Cache Merging for Memory-efficient LLMs Inference☆50Jun 19, 2024Updated last year
- Helpful kernel tutorials, examples and SKILLs for tile-based GPU programming☆727May 19, 2026Updated last week
- NVIDIA NVSHMEM is a parallel programming interface for NVIDIA GPUs based on OpenSHMEM. NVSHMEM can significantly reduce multi-process com…☆536May 5, 2026Updated 3 weeks ago
- The tool facilitates debugging convergence issues and testing new algorithms and recipes for training LLMs using Nvidia libraries such as…☆19Sep 17, 2025Updated 8 months ago