1,847 9 hours ago

Dynamically quantized only 12GB MTP model with precision close to the base model picked from hugging face ISTA-DASLab/Qwen3.8-27B-GSQ-RCO-IQ3_S-GGUF, highly efficient model

vision