diff --git a/utils.cpp b/utils.cpp index 1679ae10a..3909c974f 100644 --- a/utils.cpp +++ b/utils.cpp @@ -146,8 +146,10 @@ std::string gpt_random_prompt(std::mt19937 & rng) { // TODO: not great allocating this every time std::vector llama_tokenize(struct llama_context * ctx, const std::string & text, bool add_bos) { - std::vector res(8096); + // initialize to prompt numer of chars, since n_tokens <= n_prompt_chars + std::vector res(text.size() + (int)add_bos); int n = llama_tokenize(ctx, text.c_str(), res.data(), res.size(), add_bos); + assert(n >= 0); res.resize(n); return res;