mradermacher

/

model_requests

like 159

why nvfp4 lol? it's already quanted

and the original model quantized by us is here:
https://hf.tst.eu/model#Qwen3-Next-80B-A3B-Instruct-GGUF

the model is calibrates on italian data

!"Calibration language — calibrated on Italian data. The model retains its full multilingual capabilities, but quantization quality may be slightly optimized for Italian and similar Romance languages."

the model will be converted back to bf16 and then requanted the same way as other models, so Im not sure about the effect, as the model you provided is not a finetune, but a quantization. if you want, I can queue it, but I assume the model will be very similar to the original