Failed to load model
#1
by
cdemanet
- opened
I got "invalid character" error when I tried to load Q4 variants with LM Studio v0.2.9 and also with llama.cpp release b1663. I didn't try the Q5 variants.
I was able to run correctly using
@alvarobartt
's Q4 variants.
Hi @cdemanet indeed that’s why I created my own GGUF variants, because I had some issues with LM Studio, that’s because the K_M and K_S variants are apparently missing and AFAIK this were quantized during the transition from GGML to GGUF, not sure if it’s an actual bug / issue or only related to LM Studio, but glad that my variants helped you too 🤗