NVIDIA / Megatron-Energon
Megatron's multi-modal data loader
β171Updated this week
Alternatives and similar repositories for Megatron-Energon:
Users that are interested in Megatron-Energon are comparing it to the libraries listed below
- π Efficiently (pre)training foundation models with native PyTorch features, including FSDP for training and SDPA implementation of Flashβ¦β227Updated last week
- [ICLR 2025] COAT: Compressing Optimizer States and Activation for Memory-Efficient FP8 Trainingβ157Updated 3 weeks ago
- This repository contains the experimental PyTorch native float8 training UXβ222Updated 7 months ago
- π Collection of components for development, training, tuning, and inference of foundation models leveraging PyTorch native components.β188Updated this week
- Applied AI experiments and examples for PyTorchβ244Updated this week
- USP: Unified (a.k.a. Hybrid, 2D) Sequence Parallel Attention for Long Context Transformers Model Training and Inferenceβ445Updated 3 weeks ago
- Official repository for LightSeq: Sequence Level Parallelism for Distributed Training of Long Context Transformersβ207Updated 6 months ago
- β156Updated last month
- Zero Bubble Pipeline Parallelismβ365Updated last week
- ring-attention experimentsβ127Updated 4 months ago
- Efficient LLM Inference over Long Sequencesβ364Updated last month
- [ICLR 2025] DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Headsβ432Updated last month
- Implementation of π Ring Attention, from Liu et al. at Berkeley AI, in Pytorchβ506Updated 4 months ago
- Microsoft Automatic Mixed Precision Libraryβ579Updated 5 months ago
- [ICLR2025] Breaking Throughput-Latency Trade-off for Long Sequences with Speculative Decodingβ109Updated 3 months ago
- Ring attention implementation with flash attentionβ707Updated 2 weeks ago
- β100Updated 6 months ago
- LLM KV cache compression made easyβ435Updated this week
- Triton-based implementation of Sparse Mixture of Experts.β207Updated 3 months ago
- β143Updated last year
- The official implementation of the paper <MoA: Mixture of Sparse Attention for Automatic Large Language Model Compression>β120Updated 3 months ago
- Code for the paper "QMoE: Practical Sub-1-Bit Compression of Trillion-Parameter Models".β270Updated last year
- Skywork-MoE: A Deep Dive into Training Techniques for Mixture-of-Experts Language Modelsβ129Updated 9 months ago
- Large Context Attentionβ687Updated last month
- β179Updated 5 months ago
- Low-bit optimizers for PyTorchβ125Updated last year
- [NeurIPS 2024] KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantizationβ335Updated 7 months ago
- A tool to configure, launch and manage your machine learning experiments.β125Updated this week
- Tritonbench is a collection of PyTorch custom operators with example inputs to measure their performance.β103Updated this week