AWEQ: Post-Training Quantization with Activation-Weight Equalization for Large Language Models
This work addresses the computational and storage costs of large language models for deployment in resource-constrained environments, representing an incremental improvement over existing post-training quantization methods.
The paper tackles the challenge of balancing accuracy and hardware efficiency in quantizing large language models by introducing AWEQ, a post-training method that uses activation-weight equalization to transfer quantization difficulty from activations to weights, achieving state-of-the-art results in ultra-low-bit and W8A8 quantization on models like LLaMA and OPT.
Large language models(LLMs) exhibit excellent performance across a variety of tasks, but they come with significant computational and storage costs. Quantizing these models is an effective way to alleviate this issue. However, existing methods struggle to strike a balance between model accuracy and hardware efficiency. This is where we introduce AWEQ, a post-training method that requires no additional training overhead. AWEQ excels in both ultra-low-bit quantization and 8-bit weight and activation (W8A8) quantization. There is an observation that weight quantization is less challenging than activation quantization. AWEQ transfers the difficulty of activation quantization to weights using channel equalization, achieving a balance between the quantization difficulties of both, and thereby maximizing performance. We have further refined the equalization method to mitigate quantization bias error, ensuring the robustness of the model. Extensive experiments on popular models such as LLaMA and OPT demonstrate that AWEQ outperforms all existing post-training quantization methods for large models.