johnheo/adadim-llm

[ICLR 2024] Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models

Python