pipeline parallelism
PulseAugur coverage of pipeline parallelism — every cluster mentioning pipeline parallelism across labs, papers, and developer communities, ranked by signal.
1 day(s) with sentiment data
-
vLLM adds speculative decoding for AMD GPUs, boosting inference speed
vLLM has implemented speculative decoding for AMD GPUs, a technique that allows for faster inference by having a smaller draft model propose tokens that a larger target model then verifies. This feature, optimized for A…
-
PyTorch DDP Explained: Gradient Synchronization for Multi-GPU Training
This article provides a deep dive into PyTorch's DistributedDataParallel (DDP) for multi-GPU training. It explains the necessity of DDP due to growing model and dataset sizes, contrasting it with model parallelism. The …
-
New research shows one-step gradient delay is not a barrier for LLM pretraining
A new research paper explores asynchronous pipeline parallelism for large-scale LLM pretraining, challenging the notion that gradient delay is an insurmountable barrier. The study demonstrates that the choice of optimiz…
-
New runtime boosts pipeline-parallel AI training efficiency
Researchers have developed a new runtime system called Runtime-Readiness-First Pipeline (RRFP) designed to improve the efficiency of large-model training using pipeline parallelism. Traditional systems can suffer from i…