diff --git a/model_adapter.h b/model_adapter.h index 523b5b828..65dd7e282 100644 --- a/model_adapter.h +++ b/model_adapter.h @@ -18,7 +18,7 @@ enum FileFormat BADFORMAT=0, //unknown, uninit, or failed to load GGML=1, // 1=(original llama ggml, alpaca, GPT4ALL, GPTJ header) GGHF=2, // 2=(llama ggmf) - GGJT=3, // 3=(llama ggjt) + GGJT=3, // 3=(llama ggjt) GGJT_2=4, //newer llama format unshuffled GGJT_3=5, //using 16bit scalar @@ -54,8 +54,6 @@ enum ModelLoadResult RETRY_LOAD = 2, //used if it's suspected that the model is an older format }; -bool llama_load_model(const load_model_inputs inputs, FileFormat file_format); -generation_outputs llama_generate(const generation_inputs inputs, generation_outputs &output); ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in_file_format); generation_outputs gpttype_generate(const generation_inputs inputs, generation_outputs &output); @@ -71,5 +69,5 @@ int ArrFindIndexOf(const std::vector targetArray, const std::vector se FileFormat check_file_format(const std::string & fname); void ContextFastForward(std::vector ¤t_context_tokens, std::vector &embd_inp, - int &n_past, std::vector &last_n_tokens, const int nctx, std::vector &smartcontext, + int &n_past, std::vector &last_n_tokens, const int nctx, std::vector &smartcontext, const bool useSmartContext, const bool requireFullSubset); \ No newline at end of file