Automatically Planning Optimal Parallel Strategy for Large Language Models
This work addresses the challenge of efficiently utilizing large computing clusters for training large-scale language models, representing an incremental improvement in automated parallelization.
The paper tackles the problem of automatically planning optimal parallel strategies for large language models by proposing an algorithm that estimates training duration with 96% average accuracy and always recommends globally optimal strategies.
The number of parameters in large-scale language models based on transformers is gradually increasing, and the scale of computing clusters is also growing. The technology of quickly mobilizing large amounts of computing resources for parallel computing is becoming increasingly important. In this paper, we propose an automatic parallel algorithm that automatically plans the parallel strategy with maximum throughput based on model and hardware information. By decoupling the training time into computation, communication, and overlap, we established a training duration simulation model. Based on this simulation model, we prune the parallel solution space to shorten the search time required. The multi-node experiment results show that the algorithm can estimate the parallel training duration in real time with an average accuracy of 96%. In our test, the recommendation strategy provided by the algorithm is always globally optimal.