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

server : support reading arguments from environment variables (#9105)

Browse source 
* server : support reading arguments from environment variables

* add -fa and -dt

* readme : specify non-arg env var
This commit is contained in:
Xuan Son Nguyen 2024-08-21 11:04:34 +02:00 • committed by GitHub
parent b40eb84895
commit fc54ef0d1c
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
4 changed files with 80 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

64
common/common.cpp
View file

@ -77,6 +77,41 @@
using json = nlohmann::ordered_json;
//
// Environment variable utils
//
template<typename T>
static typename std::enable_if<std::is_same<T, std::string>::value, void>::type
get_env(std::string name, T & target) {
char * value = std::getenv(name.c_str());
target = value ? std::string(value) : target;
}
template<typename T>
static typename std::enable_if<!std::is_same<T, bool>::value && std::is_integral<T>::value, void>::type
get_env(std::string name, T & target) {
char * value = std::getenv(name.c_str());
target = value ? std::stoi(value) : target;
}
template<typename T>
static typename std::enable_if<std::is_floating_point<T>::value, void>::type
get_env(std::string name, T & target) {
char * value = std::getenv(name.c_str());
target = value ? std::stof(value) : target;
}
template<typename T>
static typename std::enable_if<std::is_same<T, bool>::value, void>::type
get_env(std::string name, T & target) {
char * value = std::getenv(name.c_str());
if (value) {
std::string val(value);
target = val == "1" || val == "true";
}
}
//
// CPU utils
//
@ -220,12 +255,6 @@ int32_t cpu_get_num_math() {
// CLI argument parsing
//
void gpt_params_handle_hf_token(gpt_params & params) {
if (params.hf_token.empty() && std::getenv("HF_TOKEN")) {
params.hf_token = std::getenv("HF_TOKEN");
}
}
void gpt_params_handle_model_default(gpt_params & params) {
if (!params.hf_repo.empty()) {
// short-hand to avoid specifying --hf-file -> default it to --model
@ -273,7 +302,9 @@ bool gpt_params_parse_ex(int argc, char ** argv, gpt_params & params) {
gpt_params_handle_model_default(params);
gpt_params_handle_hf_token(params);
if (params.hf_token.empty()) {
get_env("HF_TOKEN", params.hf_token);
}
if (params.escape) {
string_process_escapes(params.prompt);
@ -293,6 +324,25 @@ bool gpt_params_parse_ex(int argc, char ** argv, gpt_params & params) {
return true;
}
void gpt_params_parse_from_env(gpt_params & params) {
// we only care about server-related params for now
get_env("LLAMA_ARG_MODEL", params.model);
get_env("LLAMA_ARG_THREADS", params.n_threads);
get_env("LLAMA_ARG_CTX_SIZE", params.n_ctx);
get_env("LLAMA_ARG_N_PARALLEL", params.n_parallel);
get_env("LLAMA_ARG_BATCH", params.n_batch);
get_env("LLAMA_ARG_UBATCH", params.n_ubatch);
get_env("LLAMA_ARG_N_GPU_LAYERS", params.n_gpu_layers);
get_env("LLAMA_ARG_THREADS_HTTP", params.n_threads_http);
get_env("LLAMA_ARG_CHAT_TEMPLATE", params.chat_template);
get_env("LLAMA_ARG_N_PREDICT", params.n_predict);
get_env("LLAMA_ARG_ENDPOINT_METRICS", params.endpoint_metrics);
get_env("LLAMA_ARG_ENDPOINT_SLOTS", params.endpoint_slots);
get_env("LLAMA_ARG_EMBEDDINGS", params.embedding);
get_env("LLAMA_ARG_FLASH_ATTN", params.flash_attn);
get_env("LLAMA_ARG_DEFRAG_THOLD", params.defrag_thold);
}
bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
const auto params_org = params; // the example can modify the default params

2
common/common.h
View file

@ -267,7 +267,7 @@ struct gpt_params {
std::string lora_outfile = "ggml-lora-merged-f16.gguf";
};
void gpt_params_handle_hf_token(gpt_params & params);
void gpt_params_parse_from_env(gpt_params & params);
void gpt_params_handle_model_default(gpt_params & params);
bool gpt_params_parse_ex (int argc, char ** argv, gpt_params & params);