Shigangli / Ok-Topk
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.
☆23Updated last year
Related projects: ⓘ
- Hi-Speed DNN Training with Espresso: Unleashing the Full Potential of Gradient Compression with Near-Optimal Usage Strategies (EuroSys '2…☆12Updated 11 months ago
- A Cluster-Wide Model Manager to Accelerate DNN Training via Automated Training Warmup☆31Updated last year
- ☆14Updated 2 years ago
- ☆18Updated 2 years ago
- ☆19Updated last year
- ☆12Updated 3 months ago
- Compiler for Dynamic Neural Networks☆42Updated 10 months ago
- Artifact for OSDI'23: MGG: Accelerating Graph Neural Networks with Fine-grained intra-kernel Communication-Computation Pipelining on Mult…☆34Updated 6 months ago
- ☆22Updated 2 years ago
- Artifacts for our SIGCOMM'22 paper Muri☆38Updated 8 months ago
- Artifacts for our ASPLOS'23 paper ElasticFlow☆51Updated 4 months ago
- LLMServingSim: A HW/SW Co-Simulation Infrastructure for LLM Inference Serving at Scale☆32Updated last month
- SOTA Learning-augmented Systems☆32Updated 2 years ago
- ☆40Updated last year
- ☆37Updated 3 years ago
- Proteus: A High-Throughput Inference-Serving System with Accuracy Scaling☆8Updated 6 months ago
- ☆29Updated 6 months ago
- Artifact for PPoPP22 QGTC: Accelerating Quantized GNN via GPU Tensor Core.☆27Updated 2 years ago
- SHADE: Enable Fundamental Cacheability for Distributed Deep Learning Training☆28Updated last year
- ☆13Updated 2 years ago
- Cupcake: A Compression Scheduler for Scalable Communication-Efficient Distributed Training (MLSys '23)☆8Updated last year
- LLM serving cluster simulator☆55Updated 4 months ago
- ☆8Updated 2 years ago
- ☆37Updated 2 years ago
- ☆20Updated 10 months ago
- MAGIS: Memory Optimization via Coordinated Graph Transformation and Scheduling for DNN (ASPLOS'24)☆35Updated 3 months ago
- A Generic Resource-Aware Hyperparameter Tuning Execution Engine☆15Updated 2 years ago
- THC: Accelerating Distributed Deep Learning Using Tensor Homomorphic Compression☆12Updated last month
- Set of datasets for the deep learning recommendation model (DLRM).☆39Updated last year
- ☆12Updated 2 years ago