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

also support loading from llama2.c vocabulary

Browse source
This commit is contained in:
Johannes Rudolph 2023-08-10 16:32:44 +02:00
parent d2b95e7e70
commit aa26201291
No known key found for this signature in database
GPG key ID: 52AF1C9ABD77E6E5
1 changed files with 57 additions and 25 deletions
Download patch file Download diff file Expand all files Collapse all files

82
examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp
View file

@ -438,6 +438,11 @@ struct llama_file {
read_raw(&ret, sizeof(ret)); read_raw(&ret, sizeof(ret));
return ret; return ret;
} }
std::float_t read_f32() {
std::float_t ret;
read_raw(&ret, sizeof(ret));
return ret;
}
std::string read_string(std::uint32_t len) { std::string read_string(std::uint32_t len) {
std::vector<char> chars(len); std::vector<char> chars(len);
@ -491,30 +496,57 @@ void write_tensor(struct llama_file * file, struct ggml_tensor * tensor) {
file->write_raw(tensor->data, ggml_nbytes(tensor)); file->write_raw(tensor->data, ggml_nbytes(tensor));
} }
void load_vocab(const char *filename, struct llama_vocab *vocab) { bool is_ggml_file(const char *filename) {
struct llama_context_params llama_params = llama_context_default_params(); llama_file file(filename, "rb");
llama_params.vocab_only = true; if (file.size < 4) {
return false;
struct llama_model * lmodel = llama_load_model_from_file(filename, llama_params); }
struct llama_context * lctx = llama_new_context_with_model(lmodel, llama_params); uint32_t magic = file.read_u32();
return magic == LLAMA_FILE_MAGIC;
std::vector<const char *> strings; }
std::vector<float> scores;
int n_vocab = llama_n_vocab(lctx); void load_vocab(const char *filename, Config *config, struct llama_vocab *vocab) {
strings.resize(n_vocab, NULL); // heuristic to infer whether vocab is from ggml or from llama2.c vocabulary
scores.resize(n_vocab, 0); if (is_ggml_file(filename)) {
n_vocab = llama_get_vocab(lctx, strings.data(), scores.data(), n_vocab);
GGML_ASSERT(n_vocab == llama_n_vocab(lctx)); struct llama_context_params llama_params = llama_context_default_params();
vocab->id_to_token.resize(n_vocab); llama_params.vocab_only = true;
for (int i=0; i<n_vocab; ++i) {
std::string tok = std::string(strings[i]); struct llama_model * lmodel = llama_load_model_from_file(filename, llama_params);
float score = scores[i]; struct llama_context * lctx = llama_new_context_with_model(lmodel, llama_params);
vocab->id_to_token[i].tok = tok;
vocab->id_to_token[i].score = score; std::vector<const char *> strings;
vocab->token_to_id.emplace(tok, i); std::vector<float> scores;
int n_vocab = llama_n_vocab(lctx);
strings.resize(n_vocab, NULL);
scores.resize(n_vocab, 0);
n_vocab = llama_get_vocab(lctx, strings.data(), scores.data(), n_vocab);
GGML_ASSERT(n_vocab == llama_n_vocab(lctx));
vocab->id_to_token.resize(n_vocab);
for (int i=0; i<n_vocab; ++i) {
std::string tok = std::string(strings[i]);
float score = scores[i];
vocab->id_to_token[i].tok = tok;
vocab->id_to_token[i].score = score;
vocab->token_to_id.emplace(tok, i);
}
llama_free(lctx);
llama_free_model(lmodel);
} else { // assume llama2.c vocabulary
printf("Assuming llama2.c vocabulary since %s is not a ggml file\n", filename);
llama_file file(filename, "rb");
uint32_t n_vocab = config->vocab_size;
/* uint32_t max_token_length = */ file.read_u32(); // unused
vocab->id_to_token.resize(n_vocab);
for (uint32_t i=0; i<n_vocab; ++i) {
float_t score = file.read_f32();
uint32_t len = file.read_u32();
std::string tok = file.read_string(len);
vocab->id_to_token[i].tok = tok;
vocab->id_to_token[i].score = score;
vocab->token_to_id.emplace(tok, i);
}
} }
llama_free(lctx);
llama_free_model(lmodel);
} }
void stuff_karpathy_weights_into_gg(struct ggml_tensor * gg_weights, float * karpathy_weights){ void stuff_karpathy_weights_into_gg(struct ggml_tensor * gg_weights, float * karpathy_weights){
@ -684,7 +716,7 @@ void print_usage(int /*argc*/, char ** argv, const struct train_params * params)
fprintf(stderr, "\n"); fprintf(stderr, "\n");
fprintf(stderr, "options:\n"); fprintf(stderr, "options:\n");
fprintf(stderr, " -h, --help show this help message and exit\n"); fprintf(stderr, " -h, --help show this help message and exit\n");
fprintf(stderr, " --copy-vocab-from-model FNAME model path from which to copy vocab (default '%s')\n", params->fn_vocab_model); fprintf(stderr, " --copy-vocab-from-model FNAME llama2.c vocabulary or ggml model path from which to copy vocab (default '%s')\n", params->fn_vocab_model);
fprintf(stderr, " --llama2c-model FNAME [REQUIRED] model path from which to load Karpathy's llama2.c model\n"); fprintf(stderr, " --llama2c-model FNAME [REQUIRED] model path from which to load Karpathy's llama2.c model\n");
fprintf(stderr, " --llama2c-output-model FNAME model path to save the converted llama2.c model (default %s')\n", params->fn_llama2c_output_model); fprintf(stderr, " --llama2c-output-model FNAME model path to save the converted llama2.c model (default %s')\n", params->fn_llama2c_output_model);
fprintf(stderr, "\n"); fprintf(stderr, "\n");
@ -764,7 +796,7 @@ int main(int argc, char ** argv) {
} }
struct llama_vocab vocab; struct llama_vocab vocab;
load_vocab(params.fn_vocab_model, &vocab); load_vocab(params.fn_vocab_model, &config, &vocab);
struct my_llama_model model; struct my_llama_model model;
model.hparams.n_vocab = config.vocab_size; //llama_n_vocab(lctx); model.hparams.n_vocab = config.vocab_size; //llama_n_vocab(lctx);