tokenizer.vocab_size(): 32000model.embed_tokens.weight: 32001
https://github.com/ggerganov/llama.cpp/issues/442
· Sign up or log in to comment