Public/koboldcpp
1
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-09-17 08:19:49 +02:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'master' into concedo

Browse Source 
# Conflicts:
#	flake.nix
This commit is contained in:
Concedo
2023-04-26 18:01:35 +08:00
parent 27bc29128e 4afcc37869
commit 93a8e00dfa
10 changed files with 459 additions and 231 deletions
Download Patch File Download Diff File Expand all files Collapse all files
llama.cpp
+4
View File
@@ -491,6 +491,7 @@ struct llama_file_loader {
case GGML_TYPE_Q4_1:
case GGML_TYPE_Q4_2:
case GGML_TYPE_Q4_3:
case GGML_TYPE_Q8_0:
break;
default: {
throw format("unrecognized tensor type %u\n", shard.type);
@@ -565,6 +566,7 @@ struct llama_file_saver {
case GGML_TYPE_Q4_1:
case GGML_TYPE_Q4_2:
case GGML_TYPE_Q4_3:
case GGML_TYPE_Q8_0:
break;
default: LLAMA_ASSERT(false);
}
@@ -855,6 +857,7 @@ static const char *llama_ftype_name(enum llama_ftype ftype) {
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_Q8_0: return "mostly Q8_0";
default: return "unknown, may not work";
}
}
@@ -1592,6 +1595,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
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_Q8_0: quantized_type = GGML_TYPE_Q8_0; break;
default: throw format("invalid output file type %d\n", ftype);
};