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

llama : added missing model type for the smallest pythia-14m

Browse source
This commit is contained in:
Stanisław Szymczyk 2024-05-22 20:54:20 +02:00
parent d8b8ecf885
commit 1aa777f823
1 changed files with 3 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

3
llama.cpp
View file

@ -1711,6 +1711,7 @@ static llama_state g_state;
// available llama models // available llama models
enum e_model { enum e_model {
MODEL_UNKNOWN, MODEL_UNKNOWN,
MODEL_14M,
MODEL_17M, MODEL_17M,
MODEL_22M, MODEL_22M,
MODEL_33M, MODEL_33M,
@ -3793,6 +3794,7 @@ static std::string llama_model_ftype_name(llama_ftype ftype) {
static const char * llama_model_type_name(e_model type) { static const char * llama_model_type_name(e_model type) {
switch (type) { switch (type) {
case MODEL_14M: return "14M";
case MODEL_17M: return "17M"; case MODEL_17M: return "17M";
case MODEL_22M: return "22M"; case MODEL_22M: return "22M";
case MODEL_33M: return "33M"; case MODEL_33M: return "33M";
@ -4314,6 +4316,7 @@ static void llm_load_hparams(
switch (hparams.n_layer) { switch (hparams.n_layer) {
case 6: case 6:
switch (hparams.n_ff) { switch (hparams.n_ff) {
case 512: model.type = e_model::MODEL_14M; break;
case 2048: model.type = e_model::MODEL_70M; break; case 2048: model.type = e_model::MODEL_70M; break;
default: model.type = e_model::MODEL_UNKNOWN; default: model.type = e_model::MODEL_UNKNOWN;
} break; } break;