Tag
A calibration-aware Q4_K_M quantization of Qwen3.5 0.8B using SpectralQuant recovers 96.5% of the BF16 performance gap compared to the standard llama.cpp Q4_K_M quant.
SpectralQuant is a new KV cache quantization technique achieving 5.95× compression on Mistral 7B with only 7.5% perplexity overhead, significantly outperforming TurboQuant while requiring only 15 seconds of calibration per model.