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

llama : store mapped names as const char *

Browse source
This commit is contained in:
Jared Van Bortel 2024-02-02 10:42:42 -05:00
parent 1dd7aa9b1c
commit 137971b2e1
1 changed files with 9 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

18
llama.cpp
View file

@ -208,7 +208,7 @@ enum llm_arch {
LLM_ARCH_UNKNOWN, LLM_ARCH_UNKNOWN,
}; };
static std::map<llm_arch, std::string> LLM_ARCH_NAMES = { static std::map<llm_arch, const char *> LLM_ARCH_NAMES = {
{ LLM_ARCH_LLAMA, "llama" }, { LLM_ARCH_LLAMA, "llama" },
{ LLM_ARCH_FALCON, "falcon" }, { LLM_ARCH_FALCON, "falcon" },
{ LLM_ARCH_GPT2, "gpt2" }, { LLM_ARCH_GPT2, "gpt2" },
@ -285,7 +285,7 @@ enum llm_kv {
LLM_KV_TOKENIZER_RWKV, LLM_KV_TOKENIZER_RWKV,
}; };
static std::map<llm_kv, std::string> LLM_KV_NAMES = { static std::map<llm_kv, const char *> LLM_KV_NAMES = {
{ LLM_KV_GENERAL_ARCHITECTURE, "general.architecture" }, { LLM_KV_GENERAL_ARCHITECTURE, "general.architecture" },
{ LLM_KV_GENERAL_QUANTIZATION_VERSION, "general.quantization_version" }, { LLM_KV_GENERAL_QUANTIZATION_VERSION, "general.quantization_version" },
{ LLM_KV_GENERAL_ALIGNMENT, "general.alignment" }, { LLM_KV_GENERAL_ALIGNMENT, "general.alignment" },
@ -346,7 +346,7 @@ struct LLM_KV {
llm_arch arch; llm_arch arch;
std::string operator()(llm_kv kv) const { std::string operator()(llm_kv kv) const {
return ::format(LLM_KV_NAMES[kv].c_str(), LLM_ARCH_NAMES[arch].c_str()); return ::format(LLM_KV_NAMES[kv], LLM_ARCH_NAMES[arch]);
} }
}; };
@ -747,7 +747,7 @@ struct LLM_TN {
// gguf helpers // gguf helpers
// //
static std::map<int32_t, std::string> LLAMA_ROPE_SCALING_TYPES = { static std::map<int32_t, const char *> LLAMA_ROPE_SCALING_TYPES = {
{ LLAMA_ROPE_SCALING_NONE, "none" }, { LLAMA_ROPE_SCALING_NONE, "none" },
{ LLAMA_ROPE_SCALING_LINEAR, "linear" }, { LLAMA_ROPE_SCALING_LINEAR, "linear" },
{ LLAMA_ROPE_SCALING_YARN, "yarn" }, { LLAMA_ROPE_SCALING_YARN, "yarn" },
@ -2701,7 +2701,7 @@ struct llama_model_loader {
// load LLaMA models // load LLaMA models
// //
static std::string llama_model_arch_name(llm_arch arch) { static const char * llama_model_arch_name(llm_arch arch) {
auto it = LLM_ARCH_NAMES.find(arch); auto it = LLM_ARCH_NAMES.find(arch);
if (it == LLM_ARCH_NAMES.end()) { if (it == LLM_ARCH_NAMES.end()) {
return "unknown"; return "unknown";
@ -3310,11 +3310,11 @@ static void llm_load_print_meta(llama_model_loader & ml, llama_model & model) {
const auto & hparams = model.hparams; const auto & hparams = model.hparams;
const auto & vocab = model.vocab; const auto & vocab = model.vocab;
const auto rope_scaling_type = LLAMA_ROPE_SCALING_TYPES.at(hparams.rope_scaling_type_train); const char * rope_scaling_type = LLAMA_ROPE_SCALING_TYPES.at(hparams.rope_scaling_type_train);
// hparams // hparams
LLAMA_LOG_INFO("%s: format = %s\n", __func__, llama_file_version_name(ml.fver)); LLAMA_LOG_INFO("%s: format = %s\n", __func__, llama_file_version_name(ml.fver));
LLAMA_LOG_INFO("%s: arch = %s\n", __func__, LLM_ARCH_NAMES.at(model.arch).c_str()); LLAMA_LOG_INFO("%s: arch = %s\n", __func__, LLM_ARCH_NAMES.at(model.arch));
LLAMA_LOG_INFO("%s: vocab type = %s\n", __func__, llama_model_vocab_type_name(vocab.type)); LLAMA_LOG_INFO("%s: vocab type = %s\n", __func__, llama_model_vocab_type_name(vocab.type));
LLAMA_LOG_INFO("%s: n_vocab = %u\n", __func__, hparams.n_vocab); LLAMA_LOG_INFO("%s: n_vocab = %u\n", __func__, hparams.n_vocab);
LLAMA_LOG_INFO("%s: n_merges = %u\n", __func__, (int) vocab.bpe_ranks.size()); LLAMA_LOG_INFO("%s: n_merges = %u\n", __func__, (int) vocab.bpe_ranks.size());
@ -3336,7 +3336,7 @@ static void llm_load_print_meta(llama_model_loader & ml, llama_model & model) {
LLAMA_LOG_INFO("%s: n_ff = %u\n", __func__, hparams.n_ff); LLAMA_LOG_INFO("%s: n_ff = %u\n", __func__, hparams.n_ff);
LLAMA_LOG_INFO("%s: n_expert = %u\n", __func__, hparams.n_expert); LLAMA_LOG_INFO("%s: n_expert = %u\n", __func__, hparams.n_expert);
LLAMA_LOG_INFO("%s: n_expert_used = %u\n", __func__, hparams.n_expert_used); LLAMA_LOG_INFO("%s: n_expert_used = %u\n", __func__, hparams.n_expert_used);
LLAMA_LOG_INFO("%s: rope scaling = %s\n", __func__, rope_scaling_type.c_str()); LLAMA_LOG_INFO("%s: rope scaling = %s\n", __func__, rope_scaling_type);
LLAMA_LOG_INFO("%s: freq_base_train = %.1f\n", __func__, hparams.rope_freq_base_train); LLAMA_LOG_INFO("%s: freq_base_train = %.1f\n", __func__, hparams.rope_freq_base_train);
LLAMA_LOG_INFO("%s: freq_scale_train = %g\n", __func__, hparams.rope_freq_scale_train); LLAMA_LOG_INFO("%s: freq_scale_train = %g\n", __func__, hparams.rope_freq_scale_train);
LLAMA_LOG_INFO("%s: n_yarn_orig_ctx = %u\n", __func__, hparams.n_yarn_orig_ctx); LLAMA_LOG_INFO("%s: n_yarn_orig_ctx = %u\n", __func__, hparams.n_yarn_orig_ctx);
@ -10735,7 +10735,7 @@ int32_t llama_model_meta_val_str_by_index(const struct llama_model * model, int3
int32_t llama_model_desc(const struct llama_model * model, char * buf, size_t buf_size) { int32_t llama_model_desc(const struct llama_model * model, char * buf, size_t buf_size) {
return snprintf(buf, buf_size, "%s %s %s", return snprintf(buf, buf_size, "%s %s %s",
llama_model_arch_name(model->arch).c_str(), llama_model_arch_name(model->arch),
llama_model_type_name(model->type), llama_model_type_name(model->type),
llama_model_ftype_name(model->ftype).c_str()); llama_model_ftype_name(model->ftype).c_str());
} }