diff --git a/main.cpp b/main.cpp index ca0fca8b3..39c5d7b76 100644 --- a/main.cpp +++ b/main.cpp @@ -845,6 +845,8 @@ int main(int argc, char ** argv) { std::vector logits; + // Add a space in front of the first character to match OG llama tokenizer behavior + params.prompt.insert(0, 1, ' '); // tokenize the prompt std::vector embd_inp = ::llama_tokenize(vocab, params.prompt, true);