ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-02-11 08:50:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

Adding top-n-sigma sampler (#489)

Browse Source 
* Adding top-n-sigma sampler

* Fix typos in XTC PR

* Update README.md for main and server

* More README

* More README

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow
2025-06-03 17:35:09 +03:00
committed by GitHub
parent ccb265c016
commit f6d5fbdc57
9 changed files with 115 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
common/common.cpp
View File

@@ -659,6 +659,11 @@ bool gpt_params_find_arg(int argc, char ** argv, const std::string & arg, gpt_pa
sparams.xtc_threshold = std::stof(argv[i]);
return true;
}
if (arg == "--top-n-sigma") {
CHECK_ARG
sparams.top_n_sigma = std::stof(argv[i]);
return true;
}
if (arg == "--cfg-negative-prompt") {
CHECK_ARG
sparams.cfg_negative_prompt = argv[i];
@@ -1646,7 +1651,8 @@ void gpt_params_print_usage(int /*argc*/, char ** argv, const gpt_params & param
options.push_back({ "*", " --mirostat-lr N", "Mirostat learning rate, parameter eta (default: %.1f)", (double)sparams.mirostat_eta });
options.push_back({ "*", " --mirostat-ent N", "Mirostat target entropy, parameter tau (default: %.1f)", (double)sparams.mirostat_tau });
options.push_back({ "*", " --xtc-probability p", "xtc probability (default: %.1f, 0.0 = disabled)", (double)sparams.xtc_probability });
options.push_back({ "*", " --xtc-threshold t", "xtc threshold (default: %.1f, 0.0 = disabled)", (double)sparams.xtc_threshold});
options.push_back({ "*", " --xtc-threshold t", "xtc threshold (default: %.1f, >0.5 = disabled)", (double)sparams.xtc_threshold});
options.push_back({ "*", " --top-n-sigma t", "top-n-sigma parmeter (default: %.1f, 0.0 = disabled)", (double)sparams.top_n_sigma});
options.push_back({ "*", " -l TOKEN_ID(+/-)BIAS", "modifies the likelihood of token appearing in the completion,\n"
"i.e. `--logit-bias 15043+1` to increase likelihood of token ' Hello',\n"
"or `--logit-bias 15043-1` to decrease likelihood of token ' Hello'" });
@@ -3410,6 +3416,7 @@ void yaml_dump_non_result_info(FILE * stream, const gpt_params & params, const l
fprintf(stream, "mirostat_lr: %f # default: 0.1\n", sparams.mirostat_eta);
fprintf(stream, "xtc_probability: %f # default: 0.0\n", sparams.xtc_probability);
fprintf(stream, "xtc_threshold: %f # default: 0.0\n", sparams.xtc_threshold);
fprintf(stream, "top_n_sigma: %f # default: 0.0\n", sparams.top_n_sigma);
fprintf(stream, "mlock: %s # default: false\n", params.use_mlock ? "true" : "false");
fprintf(stream, "model: %s # default: %s\n", params.model.c_str(), DEFAULT_MODEL_PATH);
fprintf(stream, "model_draft: %s # default:\n", params.model_draft.c_str());