Tokenizer was broken in the PR when these were made, now fixed(?)
#1
by
Fizzarolli
- opened
see https://github.com/ggerganov/llama.cpp/pull/8156/commits/6615d09217b7fbcf7a603ce9a9181a7aadafe78f
GGUFs incl. the fp32 one probably need to be remade from scratch though :(
Fixed as of 4 minutes ago (9:44 PM EST)!
just as i get done downloading the bf16 model to make my own quants :( /lh
thanks!
Fizzarolli
changed discussion status to
closed