When Quantization Affects Confidence of Large Language Models?
This work addresses the problem of performance degradation in quantized LLMs for researchers and practitioners, though it is incremental as it builds on known issues.
The study investigated how quantization affects the confidence and calibration of Large Language Models, finding that 4-bit GPTQ quantization reduces confidence in true labels and disproportionately impacts samples where the original model had low confidence.
Recent studies introduced effective compression techniques for Large Language Models (LLMs) via post-training quantization or low-bit weight representation. Although quantized weights offer storage efficiency and allow for faster inference, existing works have indicated that quantization might compromise performance and exacerbate biases in LLMs. This study investigates the confidence and calibration of quantized models, considering factors such as language model type and scale as contributors to quantization loss. Firstly, we reveal that quantization with GPTQ to 4-bit results in a decrease in confidence regarding true labels, with varying impacts observed among different language models. Secondly, we observe fluctuations in the impact on confidence across different scales. Finally, we propose an explanation for quantization loss based on confidence levels, indicating that quantization disproportionately affects samples where the full model exhibited low confidence levels in the first place.