Don't Be Greedy, Just Relax! Pruning LLMs via Frank-Wolfe
By: Christophe Roux , Max Zimmer , Alexandre d'Aspremont and more
Potential Business Impact:
Makes big computer brains smaller and faster.
Pruning is a common technique to reduce the compute and storage requirements of Neural Networks. While conventional approaches typically retrain the model to recover pruning-induced performance degradation, state-of-the-art Large Language Model (LLM) pruning methods operate layer-wise, minimizing the per-layer pruning error on a small calibration dataset to avoid full retraining, which is considered computationally prohibitive for LLMs. However, finding the optimal pruning mask is a hard combinatorial problem and solving it to optimality is intractable. Existing methods hence rely on greedy heuristics that ignore the weight interactions in the pruning objective. In this work, we instead consider the convex relaxation of these combinatorial constraints and solve the resulting problem using the Frank-Wolfe (FW) algorithm. Our method drastically reduces the per-layer pruning error, outperforms strong baselines on state-of-the-art GPT architectures, and remains memory-efficient. We provide theoretical justification by showing that, combined with the convergence guarantees of the FW algorithm, we obtain an approximate solution to the original combinatorial problem upon rounding the relaxed solution to integrality.
Similar Papers
FastForward Pruning: Efficient LLM Pruning via Single-Step Reinforcement Learning
Machine Learning (CS)
Makes AI models smaller and faster to train.
Fewer Weights, More Problems: A Practical Attack on LLM Pruning
Machine Learning (CS)
Pruning AI models can hide bad behavior.
Projection-Free CNN Pruning via Frank-Wolfe with Momentum: Sparser Models with Less Pretraining
Machine Learning (CS)
Makes computer "brains" smaller, faster, and smarter.