Why this two are so small in file size?

#1
by faisalhr1997 - opened

minigpt4-7B-q8_1.bin
15.4 MB

minigpt4-7B-q8_k.bin
15.4 MB

Thanks for catching that. The conversion probably messed up at some point in the process.

I will update these files.

Hmm, have to look into this further, seems like Q8_1 and Q8_K quantization isn't supported in the function I'm using in ggml. It'll take a bit of time before Q8_1 and Q8_K are supported.

Sign up or log in to comment