vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Revert "main : alternative instruct mode (Vicuna support, etc.) (#863)" (#982)

Browse source 
This reverts commit f4d277ae17.
This commit is contained in:
Pavol Rusnak 2023-04-14 21:58:43 +02:00 • committed by GitHub
parent 489093548c
commit c85e03d12e
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
12 changed files with 121 additions and 445 deletions
Download patch file Download diff file Expand all files Collapse all files

30
examples/common.h
View file

@ -14,14 +14,14 @@
//
struct gpt_params {
int32_t seed = -1; // RNG seed
int32_t n_threads = std::min(4, (int32_t) std::thread::hardware_concurrency()); // max 4 threads (default)
int32_t n_predict = 128; // new tokens to predict
int32_t repeat_last_n = 64; // last n tokens to penalize
int32_t n_parts = -1; // amount of model parts (-1 = determine from model dimensions)
int32_t n_ctx = 512; // context size
int32_t n_batch = 8; // batch size for prompt processing
int32_t n_keep = 0; // number of tokens to keep from initial prompt (-1 for all)
int32_t seed = -1; // RNG seed
int32_t n_threads = std::min(4, (int32_t) std::thread::hardware_concurrency());
int32_t n_predict = 128; // new tokens to predict
int32_t repeat_last_n = 64; // last n tokens to penalize
int32_t n_parts = -1; // amount of model parts (-1 = determine from model dimensions)
int32_t n_ctx = 512; // context size
int32_t n_batch = 8; // batch size for prompt processing
int32_t n_keep = 0; // number of tokens to keep from initial prompt
// sampling parameters
int32_t top_k = 40;
@ -33,15 +33,8 @@ struct gpt_params {
std::string prompt = "";
std::string input_prefix = ""; // string to prefix user inputs with
std::string instruct_prefix = ""; // prefix user inputs with tokenized string
bool instruct_prefix_bos = false; // prepend bos token to instruct prefix
std::string instruct_suffix = ""; // suffix user inputs with tokenized string
bool instruct_suffix_bos = false; // prepend bos token to instruct suffix
std::vector<std::string> antiprompt; // string upon seeing which more user input is prompted
std::vector<std::string> stopprompt; // string upon seeing which more user input is prompted (without adding instruct prefixes and suffixes)
bool rm_trailing_space_workaround = false; // workaround for removing trailing space from reverse/stop prompts
bool memory_f16 = true; // use f16 instead of f32 for memory kv
bool random_prompt = false; // do not randomize prompt if none provided
@ -58,14 +51,11 @@ struct gpt_params {
bool use_mlock = false; // use mlock to keep model in memory
bool mem_test = false; // compute maximum memory usage
bool verbose_prompt = false; // print prompt tokens before generation
bool clean_interface = false; // hides input prefix & suffix and displays '>'
bool multiline_mode = true; // enables multi-line mode, to send input press CTRL+D on Linux/Max, Ctrl+Z then Return on Windows
};
bool gpt_params_parse(int argc, char ** argv, gpt_params & params);
void gpt_print_usage(char * argv_0, const gpt_params & params);
void gpt_print_usage(int argc, char ** argv, const gpt_params & params);
std::string gpt_random_prompt(std::mt19937 & rng);
@ -105,5 +95,3 @@ void set_console_color(console_state & con_st, console_color_t color);
void win32_console_init(bool enable_color);
void win32_utf8_encode(const std::wstring & wstr, std::string & str);
#endif
bool get_input_text(std::string & input_text, bool escape_newline_mode);