vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-10 00:54:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

recognize q8_0 as an older format as the new clblast doesnt work correctly with it

Browse source
This commit is contained in:
Concedo 2023-05-14 11:06:23 +08:00
parent 49d6334dc1
commit 8a5fe628df
2 changed files with 13 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

4
gpttype_adapter.cpp
View file

@ -228,7 +228,7 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
if(file_format == FileFormat::GGML || file_format == FileFormat::GGHF || file_format == FileFormat::GGJT || file_format == FileFormat::GGJT_2)
{
//newer format has bit unshuffling
SetQuantsUnshuffled(file_format== FileFormat::GGJT_2);
SetQuantsUnshuffled(file_format == FileFormat::GGJT_2);
llama_ctx_params = llama_context_default_params();
llama_ctx_params.n_ctx = inputs.max_context_length;
@ -248,7 +248,7 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
}
if (file_format < FileFormat::GGJT_2)
{
printf("\n---\nWarning: Your model has an INVALID or OUTDATED format (ver %d). Please reconvert it for better results!\n---\n", file_format);
printf("\n---\nWarning: Your model may be an OUTDATED format (ver %d). Please reconvert it for better results!\n---\n", file_format);
}
if (lora_filename != "")