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

imatrix : migrate to gpt_params (#7771)

Browse source 
* imatrix : migrate to gpt_params

ggml-ci

* imatrix : add --save-frequency cli arg

* common : fix --no-ppl
This commit is contained in:
Georgi Gerganov 2024-06-06 16:30:58 +03:00 • committed by GitHub
parent ad675e1c67
commit f83351f9a6
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 213 additions and 215 deletions
Download patch file Download diff file Expand all files Collapse all files

75
common/common.cpp
View file

@ -273,6 +273,7 @@ bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
}
} catch (const std::invalid_argument & ex) {
fprintf(stderr, "%s\n", ex.what());
params = params_org;
return false;
}
@ -408,6 +409,20 @@ bool gpt_params_find_arg(int argc, char ** argv, const std::string & arg, gpt_pa
}
return true;
}
if (arg == "--in-file") {
if (++i >= argc) {
invalid_param = true;
return true;
}
std::ifstream file(argv[i]);
if (!file) {
fprintf(stderr, "error: failed to open file '%s'\n", argv[i]);
invalid_param = true;
return true;
}
params.in_files.push_back(argv[i]);
return true;
}
if (arg == "-n" || arg == "--predict" || arg == "--n-predict") {
if (++i >= argc) {
invalid_param = true;
@ -1081,7 +1096,15 @@ bool gpt_params_find_arg(int argc, char ** argv, const std::string & arg, gpt_pa
return true;
}
if (arg == "-v" || arg == "--verbose") {
params.verbose = true;
params.verbosity = 1;
return true;
}
if (arg == "--verbosity") {
if (++i >= argc) {
invalid_param = true;
return true;
}
params.verbosity = std::stoi(argv[i]);
return true;
}
if (arg == "--verbose-prompt") {
@ -1537,6 +1560,46 @@ bool gpt_params_find_arg(int argc, char ** argv, const std::string & arg, gpt_pa
params.i_pos = std::stoi(argv[i]);
return true;
}
if (arg == "-o" || arg == "--output" || arg == "--output-file") {
if (++i >= argc) {
invalid_param = true;
return true;
}
params.out_file = argv[i];
return true;
}
if (arg == "-ofreq" || arg == "--output-frequency") {
if (++i >= argc) {
invalid_param = true;
return true;
}
params.n_out_freq = std::stoi(argv[i]);
return true;
}
if (arg == "--save-frequency") {
if (++i >= argc) {
invalid_param = true;
return true;
}
params.n_save_freq = std::stoi(argv[i]);
return true;
}
if (arg == "--process-output") {
params.process_output = true;
return true;
}
if (arg == "--no-ppl") {
params.compute_ppl = false;
return true;
}
if (arg == "--chunk" || arg == "--from-chunk") {
if (++i >= argc) {
invalid_param = true;
return true;
}
params.i_chunk = std::stoi(argv[i]);
return true;
}
#ifndef LOG_DISABLE_LOGS
// Parse args for logging parameters
if (log_param_single_parse(argv[i])) {
@ -1612,6 +1675,7 @@ void gpt_params_print_usage(int /*argc*/, char ** argv, const gpt_params & param
options.push_back({ "*", "-h, --help, --usage", "print usage and exit" });
options.push_back({ "*", " --version", "show version and build info" });
options.push_back({ "*", "-v, --verbose", "print verbose information" });
options.push_back({ "*", " --verbosity N", "set specific verbosity level (default: %d)", params.verbosity });
options.push_back({ "*", " --verbose-prompt", "print a verbose prompt before generation (default: %s)", params.verbose_prompt ? "true" : "false" });
options.push_back({ "*", " --no-display-prompt", "don't print prompt at generation (default: %s)", !params.display_prompt ? "true" : "false" });
options.push_back({ "*", "-co, --color", "colorise output to distinguish prompt and user input from generations (default: %s)", params.use_color ? "true" : "false" });
@ -1637,6 +1701,7 @@ void gpt_params_print_usage(int /*argc*/, char ** argv, const gpt_params & param
options.push_back({ "*", "-fa, --flash-attn", "enable Flash Attention (default: %s)", params.flash_attn ? "enabled" : "disabled" });
options.push_back({ "*", "-p, --prompt PROMPT", "prompt to start generation with (default: '%s')", params.prompt.c_str() });
options.push_back({ "*", "-f, --file FNAME", "a file containing the prompt (default: none)" });
options.push_back({ "*", " --in-file FNAME", "an input file (repeat to specify multiple files)" });
options.push_back({ "*", "-bf, --binary-file FNAME", "binary file containing the prompt (default: none)" });
options.push_back({ "*", "-e, --escape", "process escapes sequences (\\n, \\r, \\t, \\', \\\", \\\\) (default: %s)", params.escape ? "true" : "false" });
options.push_back({ "*", " --no-escape", "do not process escape sequences" });
@ -1804,6 +1869,14 @@ void gpt_params_print_usage(int /*argc*/, char ** argv, const gpt_params & param
options.push_back({ "passkey", " --junk N", "number of times to repeat the junk text (default: %d)", params.n_junk });
options.push_back({ "passkey", " --pos N", "position of the passkey in the junk text (default: %d)", params.i_pos });
options.push_back({ "imatrix" });
options.push_back({ "imatrix", "-o, --output FNAME", "output file (default: '%s')", params.out_file.c_str() });
options.push_back({ "imatrix", " --output-frequency N", "output the imatrix every N iterations (default: %d)", params.n_out_freq });
options.push_back({ "imatrix", " --save-frequency N", "save an imatrix copy every N iterations (default: %d)", params.n_save_freq });
options.push_back({ "imatrix", " --process-output", "collect data for the output tensor (default: %s)", params.process_output ? "true" : "false" });
options.push_back({ "imatrix", " --no-ppl", "do not compute perplexity (default: %s)", params.compute_ppl ? "true" : "false" });
options.push_back({ "imatrix", " --chunk N", "start processing the input from chunk N (default: %d)", params.i_chunk });
options.push_back({ "bench" });
options.push_back({ "bench", "-pps", "is the prompt shared across parallel sequences (default: %s)", params.is_pp_shared ? "true" : "false" });
options.push_back({ "bench", "-npp n0,n1,...", "number of prompt tokens" });