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

Remove Q4_3 which is no better than Q5 (#1218)

Browse source
This commit is contained in:
Stephan Walter 2023-04-28 23:10:43 +00:00 • committed by GitHub
parent 7f15c5c477
commit 36d19a603b
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
11 changed files with 21 additions and 359 deletions
Download patch file Download diff file Expand all files Collapse all files

4
llama.cpp
View file

@ -483,7 +483,6 @@ struct llama_file_loader {
case GGML_TYPE_Q4_0:
case GGML_TYPE_Q4_1:
case GGML_TYPE_Q4_2:
case GGML_TYPE_Q4_3:
case GGML_TYPE_Q5_0:
case GGML_TYPE_Q5_1:
case GGML_TYPE_Q8_0:
@ -560,7 +559,6 @@ struct llama_file_saver {
case GGML_TYPE_Q4_0:
case GGML_TYPE_Q4_1:
case GGML_TYPE_Q4_2:
case GGML_TYPE_Q4_3:
case GGML_TYPE_Q5_0:
case GGML_TYPE_Q5_1:
case GGML_TYPE_Q8_0:
@ -853,7 +851,6 @@ static const char *llama_ftype_name(enum llama_ftype ftype) {
case LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16:
return "mostly Q4_1, some F16";
case LLAMA_FTYPE_MOSTLY_Q4_2: return "mostly Q4_2";
case LLAMA_FTYPE_MOSTLY_Q4_3: return "mostly Q4_3";
case LLAMA_FTYPE_MOSTLY_Q5_0: return "mostly Q5_0";
case LLAMA_FTYPE_MOSTLY_Q5_1: return "mostly Q5_1";
case LLAMA_FTYPE_MOSTLY_Q8_0: return "mostly Q8_0";
@ -1593,7 +1590,6 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
case LLAMA_FTYPE_MOSTLY_Q4_0: quantized_type = GGML_TYPE_Q4_0; break;
case LLAMA_FTYPE_MOSTLY_Q4_1: quantized_type = GGML_TYPE_Q4_1; break;
case LLAMA_FTYPE_MOSTLY_Q4_2: quantized_type = GGML_TYPE_Q4_2; break;
case LLAMA_FTYPE_MOSTLY_Q4_3: quantized_type = GGML_TYPE_Q4_3; break;
case LLAMA_FTYPE_MOSTLY_Q5_0: quantized_type = GGML_TYPE_Q5_0; break;
case LLAMA_FTYPE_MOSTLY_Q5_1: quantized_type = GGML_TYPE_Q5_1; break;
case LLAMA_FTYPE_MOSTLY_Q8_0: quantized_type = GGML_TYPE_Q8_0; break;