leafspark

/

Meta-Llama-3.1-405B-Instruct-GGUF

like 22

Hi I have been following the discussion at https://github.com/ggerganov/llama.cpp/pull/8676 and https://github.com/ggerganov/llama.cpp/issues/8650 and Smaug BPE is NOT the same as Llama 3.1's tokenizer. Have you tested these quants at 110k ctx?