Ok-Topk is a scheme for distributed training with sparse gradients. Ok-Topk integrates a novel sparse allreduce algorithm (less than 6k communication volume which is asymptotically optimal) with the decentralized parallel Stochastic Gradient Descent (SGD) optimizer, and its convergence is proved theoretically and empirically.
☆27Dec 10, 2022Updated 3 years ago
Alternatives and similar repositories for Ok-Topk
Users that are interested in Ok-Topk are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- A Sparse-tensor Communication Framework for Distributed Deep Learning☆13Nov 1, 2021Updated 4 years ago
- [PACT'24] GraNNDis. A fast and unified distributed graph neural network (GNN) training framework for both full-batch (full-graph) and min…☆10Aug 13, 2024Updated 2 years ago
- We present a set of all-reduce compatible gradient compression algorithms which significantly reduce the communication overhead while mai…☆10Nov 14, 2021Updated 4 years ago
- Chimera: bidirectional pipeline parallelism for efficiently training large-scale models.☆72Mar 20, 2025Updated last year
- Artifacts of VLDB'22 paper "COMET: A Novel Memory-Efficient Deep Learning TrainingFramework by Using Error-Bounded Lossy Compression"☆10Aug 2, 2022Updated 4 years ago
- 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.
- Practical low-rank gradient compression for distributed optimization: https://arxiv.org/abs/1905.13727☆151Oct 29, 2024Updated last year
- Understanding Top-k Sparsification in Distributed Deep Learning☆24Nov 15, 2019Updated 6 years ago
- Hi-Speed DNN Training with Espresso: Unleashing the Full Potential of Gradient Compression with Near-Optimal Usage Strategies (EuroSys '2…☆15Sep 21, 2023Updated 3 years ago
- A computation-parallel deep learning architecture.☆13Sep 25, 2019Updated 6 years ago
- RPCNIC: A High-Performance and Reconfigurable PCIe-attached RPC Accelerator [HPCA2025]☆16Dec 9, 2024Updated last year
- Magicube is a high-performance library for quantized sparse matrix operations (SpMM and SDDMM) of deep learning on Tensor Cores.☆92Nov 23, 2022Updated 3 years ago
- ☆17Sep 25, 2025Updated 11 months ago
- GRACE - GRAdient ComprEssion for distributed deep learning☆141Jul 23, 2024Updated 2 years ago
- C++17 implementation of einops for libtorch - clear and reliable tensor manipulations with einstein-like notation☆12Oct 16, 2023Updated 2 years ago
- 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.
- A compressed adaptive optimizer for training large-scale deep learning models using PyTorch☆25Nov 26, 2019Updated 6 years ago
- ☆15Nov 7, 2025Updated 10 months ago
- ☆12Dec 26, 2024Updated last year
- ☆69Mar 14, 2023Updated 3 years ago
- ☆10Jun 4, 2021Updated 5 years ago
- Code for reproducing experiments performed for Accoridon☆13Jun 11, 2021Updated 5 years ago
- Sample Codes using NVSHMEM on Multi-GPU☆30Jan 22, 2023Updated 3 years ago
- [DATE 2023] Pipe-BD: Pipelined Parallel Blockwise Distillation☆12Jul 13, 2023Updated 3 years ago
- Take your first step in writing a compiler. Implemented in Rust.☆17Apr 17, 2023Updated 3 years ago
- Managed Database hosting by DigitalOcean • AdPostgreSQL, MySQL, MongoDB, Kafka, Valkey, and OpenSearch available. Automatically scale up storage and focus on building your apps.
- Recycling Model Updates in Federated Learning: Are Gradient Subspaces Low-Rank?☆15Mar 24, 2022Updated 4 years ago
- gTop-k S-SGD: A Communication-Efficient Distributed Synchronous SGD Algorithm for Deep Learning☆37Aug 19, 2019Updated 7 years ago
- ☆21Jun 29, 2022Updated 4 years ago
- Sparsified SGD with Memory: https://arxiv.org/abs/1809.07599☆58Oct 25, 2018Updated 7 years ago
- Official Implementation of "RTop-K: Ultra-Fast Row-Wise Top-K Selection for Neural Network Acceleration on GPUs"☆29Jul 23, 2025Updated last year
- ☆22Aug 13, 2024Updated 2 years ago
- [ICDCS 2023] DeAR: Accelerating Distributed Deep Learning with Fine-Grained All-Reduce Pipelining☆12Dec 4, 2023Updated 2 years ago
- Paper list of federated learning: About system design☆13Apr 13, 2022Updated 4 years ago
- [HPCA'24] Smart-Infinity: Fast Large Language Model Training using Near-Storage Processing on a Real System☆52Jul 21, 2025Updated last year
- 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.
- ☆13Nov 25, 2022Updated 3 years ago
- ☆17Jun 25, 2024Updated 2 years ago
- Save your network configuration in a bash script☆22Jan 29, 2024Updated 2 years ago
- Partial implementation of paper "DEEP GRADIENT COMPRESSION: REDUCING THE COMMUNICATION BANDWIDTH FOR DISTRIBUTED TRAINING"☆33Nov 20, 2020Updated 5 years ago
- A rust-based benchmark for BlueField SmartNICs.☆30Jul 5, 2023Updated 3 years ago
- InstAttention: In-Storage Attention Offloading for Cost-Effective Long-Context LLM Inference☆19Mar 30, 2025Updated last year
- ☆11Apr 29, 2023Updated 3 years ago