refactor gpt_params_parse

common/arg.cpp

@@ -169,7 +169,7 @@ static void gpt_params_handle_model_default(gpt_params & params) {
 // CLI argument parsing functions
 //
 
-static bool gpt_params_parse_ex(int argc, char ** argv, llama_arg_context & ctx_arg) {
+static bool gpt_params_parse_ex(int argc, char ** argv, gpt_params_context & ctx_arg) {
     std::string arg;
     const std::string arg_prefix = "--";
     gpt_params & params = ctx_arg.params;
@@ -290,7 +290,7 @@ static bool gpt_params_parse_ex(int argc, char ** argv, llama_arg_context & ctx_
     return true;
 }
 
-static void gpt_params_print_usage(llama_arg_context & ctx_arg) {
+static void gpt_params_print_usage(gpt_params_context & ctx_arg) {
     auto print_options = [](std::vector<llama_arg *> & options) {
         for (llama_arg * opt : options) {
             printf("%s", opt->to_string().c_str());
@@ -319,7 +319,8 @@ static void gpt_params_print_usage(llama_arg_context & ctx_arg) {
     print_options(specific_options);
 }
 
-bool gpt_params_parse(int argc, char ** argv, llama_arg_context & ctx_arg) {
+bool gpt_params_parse(int argc, char ** argv, gpt_params & params, llama_example ex, void(*print_usage)(int, char **)) {
+    auto ctx_arg = gpt_params_parser_init(params, ex, print_usage);
     const gpt_params params_org = ctx_arg.params; // the example can modify the default params
 
     try {
@@ -343,8 +344,8 @@ bool gpt_params_parse(int argc, char ** argv, llama_arg_context & ctx_arg) {
     return true;
 }
 
-llama_arg_context gpt_params_parser_init(gpt_params & params, llama_example ex, void(*print_usage)(int, char **)) {
+gpt_params_context gpt_params_parser_init(gpt_params & params, llama_example ex, void(*print_usage)(int, char **)) {
-    llama_arg_context ctx_arg(params);
+    gpt_params_context ctx_arg(params);
     ctx_arg.print_usage = print_usage;
     ctx_arg.ex          = ex;
 

common/arg.h

@@ -61,17 +61,17 @@ struct llama_arg {
     std::string to_string();
 };
 
-struct llama_arg_context {
+struct gpt_params_context {
     enum llama_example ex = LLAMA_EXAMPLE_COMMON;
     gpt_params & params;
     std::vector<llama_arg> options;
     void(*print_usage)(int, char **) = nullptr;
-    llama_arg_context(gpt_params & params) : params(params) {}
+    gpt_params_context(gpt_params & params) : params(params) {}
 };
 
-// optionally, we can provide "print_usage" to print example usage
-llama_arg_context gpt_params_parser_init(gpt_params & params, llama_example ex, void(*print_usage)(int, char **) = nullptr);
-
 // parse input arguments from CLI
 // if one argument has invalid value, it will automatically display usage of the specific argument (and not the full usage message)
-bool gpt_params_parse(int argc, char ** argv, llama_arg_context & ctx_arg);
+bool gpt_params_parse(int argc, char ** argv, gpt_params & params, llama_example ex, void(*print_usage)(int, char **) = nullptr);
+
+// function to be used by test-arg-parser
+gpt_params_context gpt_params_parser_init(gpt_params & params, llama_example ex, void(*print_usage)(int, char **) = nullptr);

examples/batched-bench/batched-bench.cpp

@@ -38,8 +38,7 @@ static void print_usage(int, char ** argv) {
 int main(int argc, char ** argv) {
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_BENCH, print_usage);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_BENCH, print_usage)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/batched/batched.cpp

@@ -19,8 +19,7 @@ int main(int argc, char ** argv) {
     params.prompt = "Hello my name is";
     params.n_predict = 32;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_COMMON, print_usage);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMMON, print_usage)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/cvector-generator/cvector-generator.cpp

@@ -389,8 +389,7 @@ static int prepare_entries(gpt_params & params, train_context & ctx_train) {
 int main(int argc, char ** argv) {
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_CVECTOR_GENERATOR, print_usage);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_CVECTOR_GENERATOR, print_usage)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/embedding/embedding.cpp

@@ -80,8 +80,7 @@ static void batch_decode(llama_context * ctx, llama_batch & batch, float * outpu
 int main(int argc, char ** argv) {
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_EMBEDDING);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_EMBEDDING)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/eval-callback/eval-callback.cpp

@@ -145,8 +145,7 @@ int main(int argc, char ** argv) {
 
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_COMMON);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMMON)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/export-lora/export-lora.cpp

@@ -402,8 +402,7 @@ static void print_usage(int, char ** argv) {
 int main(int argc, char ** argv) {
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_EXPORT_LORA, print_usage);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_EXPORT_LORA, print_usage)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/gen-docs/gen-docs.cpp

@@ -10,7 +10,7 @@ static void export_md(std::string fname, llama_example ex) {
     std::ofstream file(fname, std::ofstream::out | std::ofstream::trunc);
 
     gpt_params params;
-    auto ctx_arg = gpt_params_parser_init(params, ex);
+    if (!gpt_params_parse(argc, argv, params, ex)) {
 
     file << "| Argument | Explanation |\n";
     file << "| -------- | ----------- |\n";

examples/gritlm/gritlm.cpp

@@ -155,8 +155,7 @@ static std::string gritlm_instruction(const std::string & instruction) {
 int main(int argc, char * argv[]) {
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_COMMON);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMMON)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/imatrix/imatrix.cpp

@@ -578,8 +578,7 @@ int main(int argc, char ** argv) {
     params.logits_all = true;
     params.verbosity = 1;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_IMATRIX, print_usage);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_IMATRIX, print_usage)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/infill/infill.cpp

@@ -106,8 +106,7 @@ int main(int argc, char ** argv) {
     gpt_params params;
     g_params = &params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_INFILL);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_INFILL)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/llava/llava-cli.cpp

@@ -279,8 +279,7 @@ int main(int argc, char ** argv) {
 
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_LLAVA, print_usage);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_LLAVA, print_usage)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/llava/minicpmv-cli.cpp

@@ -255,8 +255,7 @@ int main(int argc, char ** argv) {
 
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_COMMON, show_additional_info);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMMON, show_additional_info)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/lookahead/lookahead.cpp

@@ -38,8 +38,7 @@ struct ngram_container {
 int main(int argc, char ** argv) {
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_COMMON);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMMON)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/lookup/lookup-create.cpp

@@ -14,8 +14,7 @@
 int main(int argc, char ** argv){
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_LOOKUP);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_LOOKUP)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/lookup/lookup-stats.cpp

@@ -16,8 +16,7 @@
 int main(int argc, char ** argv){
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_LOOKUP);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_LOOKUP)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/lookup/lookup.cpp

@@ -14,8 +14,7 @@
 int main(int argc, char ** argv){
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_LOOKUP);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_LOOKUP)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/main/main.cpp

@@ -139,8 +139,7 @@ static std::string chat_add_and_format(struct llama_model * model, std::vector<l
 int main(int argc, char ** argv) {
     gpt_params params;
     g_params = &params;
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_MAIN, print_usage);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_MAIN, print_usage)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/parallel/parallel.cpp

@@ -102,8 +102,7 @@ int main(int argc, char ** argv) {
 
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_PARALLEL);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_PARALLEL)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/passkey/passkey.cpp

@@ -20,8 +20,7 @@ int main(int argc, char ** argv) {
     params.n_keep = 32;
     params.i_pos  = -1;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_PASSKEY, print_usage);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_PASSKEY, print_usage)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/perplexity/perplexity.cpp

@@ -1968,8 +1968,7 @@ int main(int argc, char ** argv) {
     params.n_ctx = 512;
     params.logits_all = true;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_PERPLEXITY);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_PERPLEXITY)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/retrieval/retrieval.cpp

@@ -112,8 +112,7 @@ static void batch_decode(llama_context * ctx, llama_batch & batch, float * outpu
 int main(int argc, char ** argv) {
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_RETRIEVAL, print_usage);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_RETRIEVAL, print_usage)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/save-load-state/save-load-state.cpp

@@ -11,8 +11,7 @@ int main(int argc, char ** argv) {
     params.prompt = "The quick brown fox";
     params.sparams.seed = 1234;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_COMMON);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMMON)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/server/server.cpp

@@ -2425,8 +2425,7 @@ int main(int argc, char ** argv) {
     // own arguments required by this example
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_SERVER);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_SERVER)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/simple/simple.cpp

@@ -19,8 +19,7 @@ int main(int argc, char ** argv) {
     params.prompt = "Hello my name is";
     params.n_predict = 32;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_COMMON, print_usage);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMMON, print_usage)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

examples/speculative/speculative.cpp

@@ -29,8 +29,7 @@ struct seq_draft {
 int main(int argc, char ** argv) {
     gpt_params params;
 
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_SPECULATIVE);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_SPECULATIVE)) {
-    if (!gpt_params_parse(argc, argv, ctx_arg)) {
         return 1;
     }
 

tests/test-arg-parser.cpp

@@ -53,7 +53,7 @@ int main(void) {
     };
 
     std::vector<std::string> argv;
-    auto ctx_arg = gpt_params_parser_init(params, LLAMA_EXAMPLE_COMMON);
+    if (!gpt_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMMON)) {
 
     printf("test-arg-parser: test invalid usage\n\n");