Is `model.gguf` t5-small or flan-t5-small?

#1
by Felladrin - opened

Hi @lmz ,

I would appreciate it if you could clarify whether model.gguf is related to t5-small or flan-t5-small.
Asking because this repository has a config-flan-t5-small.json without a corresponding model-flan-t5-small.gguf, and by @radames ' space (https://huggingface.co/spaces/radames/Candle-T5-Generation-Wasm), I'd guess the file is related to t5-small:

image.png

See also: Code reference

If that's the case, could you upload a model-flan-t5-small.gguf?

It's indeed t5-small, you can see where the mapping is done in our command line example here. This was the first model to be quantized hence the generic name.

Thanks for confirming! So I'm going to generate flan-t5-small.gguf following those instructions from the readme.

Felladrin changed discussion status to closed

Sign up or log in to comment