Fetching the paper…

SpQR: A Sparse-Quantized Representation for Near-Lossless LLM Weight Compression · Around