cahya/gpt2-large-indonesian-522M Quant?

#83
by Spectre5390 - opened

That model is already quantized - llama.cpp can only work with uqnuantized models unfortunately, i.e. it would need to be a pure f16/bf16/f32 version.

I tried it anyway to see how it would bail out and have more bad news: The (pre-)tokenizer isn't supported by llama.cpp either.

mradermacher changed discussion status to closed

Dang, that's too bad. Thanks for trying.

Sure - don't hold back on future requests because of this setback :)

Sign up or log in to comment