What type of study is this?

This is a Experimental Study study.

October 3, 20250 citationsOpen Access

ICQuant Framework for Low-Bit Weight Quantization in Language Models

ICQuant: Index Coding enables Low-bit LLM Quantization

View Full Paper

Ask AI

Bookmark

Authors

ICQuant demonstrates improved outlier suppression in low-bit quantization, enhancing model accuracy significantly.

ICQuant improves quantization quality while reducing bit overhead to 0.3 bits, a significant saving.
Using ICQuant, zero-shot accuracy of the Llama3-70B model improves by up to 150% relative to existing methods.
The framework efficiently targets outliers, minimizing their impact on quantization range and errors.
ICQuant can be integrated with existing quantizers, supporting better performance without requiring extensive fine-tuning.

Li et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

October 3, 20250 citationsOpen Access

ICQuant: Index Coding enables Low-bit LLM Quantization

View Full Paper

Ask AI

Bookmark

Authors

ICQuant demonstrates improved outlier suppression in low-bit quantization, enhancing model accuracy significantly.

ICQuant improves quantization quality while reducing bit overhead to 0.3 bits, a significant saving.
Using ICQuant, zero-shot accuracy of the Llama3-70B model improves by up to 150% relative to existing methods.
The framework efficiently targets outliers, minimizing their impact on quantization range and errors.
ICQuant can be integrated with existing quantizers, supporting better performance without requiring extensive fine-tuning.

Li et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark