Skip to yearly menu bar Skip to main content


Poster

ExCP: Extreme LLM Checkpoint Compression via Weight-Momentum Joint Shrinking

Wenshuo Li · Xinghao Chen · Han Shu · Yehui Tang · Yunhe Wang

Hall C 4-9
[ ]
Wed 24 Jul 4:30 a.m. PDT — 6 a.m. PDT
 
Oral presentation: Oral 4x LLMs
Wed 24 Jul 7:30 a.m. PDT — 8:30 a.m. PDT

Abstract: Large language models (LLM) have recently attracted significant attention in the field of artificial intelligence. However, the training process of these models poses significant challenges in terms of computational and storage capacities, thus compressing checkpoints has become an urgent problem. In this paper, we propose a novel Extreme Checkpoint Compression (ExCP) framework, which significantly reduces the required storage of training checkpoints while achieving nearly lossless performance. We first calculate the residuals of adjacent checkpoints to obtain the essential but sparse information for higher compression ratio. To further excavate the redundancy parameters in checkpoints, we then propose a weight-momentum joint shrinking method to utilize another important information during the model optimization, i.e., momentum. In particular, we exploit the information of both model and optimizer to discard as many parameters as possible while preserving critical information to ensure optimal performance. Furthermore, we utilize non-uniform quantization to further compress the storage of checkpoints. We extensively evaluate our proposed ExCP framework on several models ranging from 410M to 7B parameters and demonstrate significant storage reduction while maintaining strong performance. For instance, we achieve approximately $70\times$ compression for the Pythia-410M model, with the final performance being as accurate as the original model on various downstream tasks. Codes will be available at https://github.com/Gaffey/ExCP.

Live content is unavailable. Log in and register to view live content