ed
/
rwkv.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Restore Q4_1_O support

This commit is contained in:
saharNooby 2023-04-17 18:53:48 +04:00
parent 28e354c183
commit 38eea116b8
1 changed files with 3 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
rwkv.cpp
View File

@ -48,8 +48,7 @@ static const ggml_type FORMAT_TYPE_TO_GGML_TYPE[5] = {
GGML_TYPE_F16, GGML_TYPE_F16,
GGML_TYPE_Q4_0, GGML_TYPE_Q4_0,
GGML_TYPE_Q4_1, GGML_TYPE_Q4_1,
// TODO Restore GGML_TYPE_Q4_1_O
//GGML_TYPE_Q4_1_O
}; };
// --- Model definition and loading utilities --- // --- Model definition and loading utilities ---
@ -756,11 +755,10 @@ bool rwkv_quantize_model_file(const char * model_file_path_in, const char * mode
{ {
cur_size = ggml_quantize_q4_1(data_f32.data(), work.data(), nelements, ne[0], hist_cur.data()); cur_size = ggml_quantize_q4_1(data_f32.data(), work.data(), nelements, ne[0], hist_cur.data());
} break; } break;
// TODO Restore case GGML_TYPE_Q4_1_O:
/*case GGML_TYPE_Q4_1_O:
{ {
cur_size = ggml_quantize_q4_1_o(data_f32.data(), work.data(), nelements, ne[0], hist_cur.data()); cur_size = ggml_quantize_q4_1_o(data_f32.data(), work.data(), nelements, ne[0], hist_cur.data());
} break;*/ } break;
default: default:
{ {
fprintf(stderr, "unsupported quantization type %d\n", type); fprintf(stderr, "unsupported quantization type %d\n", type);