Hacker News Reader

This strikes me as less a leak and more clever marketing from Mistral.

That isn't unquantized, it's de-quantized. They went from Q5 to fp16 for use in Pytorch instead of the GGUF ecosystem.

I never thought people would be upscaling models by increasing quantization precision. The rationale makes sense bit its also a goofy outcome.

You should be able to upscale and fine tune to recover performance, I suppose!

Clearly we should train a diffusion model to denoise the weights of LLM transformer models. Yo dawg.

Yes, that’s correct. Good correction.