Hello. Thank you for the nice model. Could you upload tokenizer.model file, please?
#7
by
NikolayKozloff
- opened
I tried to convert your model to ggml using python script from llama.cpp github repo but it raised an error "could not find tokenizer.model".
Hi,
we don't have this file at the moment since we fine-tuned the Hugging Face version of Llama 2. You may use the one in meta-llama/Llama-2-13b, but since we used a padding token the generated responses will be different/worse.