A Convex-optimization-based Layer-wise Post-training Pruner for Large Language Models
This addresses the challenge of compressing LLMs for memory and computational savings, which is crucial for deploying these models in resource-constrained environments, representing a novel method rather than an incremental improvement.
The paper tackles the problem of pruning large language models (LLMs) efficiently without retraining, introducing FISTAPruner, a convex-optimization-based post-training pruner that achieves superior performance over state-of-the-art methods on models like OPT and LLaMA with up to 70B parameters.
Pruning is a critical strategy for compressing trained large language models (LLMs), aiming at substantial memory conservation and computational acceleration without compromising performance. However, existing pruning methods often necessitate inefficient retraining for billion-scale LLMs or rely on heuristic methods such as the optimal brain surgeon framework, which degrade performance. In this paper, we introduce FISTAPruner, the first post-training pruner based on convex optimization models and algorithms. Specifically, we propose a convex optimization model incorporating $\ell_1$ norm to induce sparsity and utilize the FISTA solver for optimization. FISTAPruner incorporates an intra-layer cumulative error correction mechanism and supports parallel pruning. We comprehensively evaluate FISTAPruner on models such as OPT, LLaMA, LLaMA-2, and LLaMA-3 with 125M to 70B parameters under unstructured and 2:4 semi-structured sparsity, demonstrating superior performance over existing state-of-the-art methods across various language benchmarks.