Public/koboldcpp
1
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-09-19 01:05:09 +02:00
Code Issues Packages Projects Releases Wiki Activity

arranged files, updated kobold lite, modified makefile for extra link args on linux, started RWKV implementation

Browse Source
This commit is contained in:
Concedo
2023-04-17 17:31:45 +08:00
parent 9581171a9f
commit 763ad172c0
21 changed files with 13597 additions and 46 deletions
Download Patch File Download Diff File Expand all files Collapse all files
expose.cpp
+31 -11
View File
@@ -53,16 +53,23 @@ extern "C"
ModelLoadResult lr = gpttype_load_model(inputs, file_format);
if (lr == ModelLoadResult::RETRY_LOAD)
{
file_format = FileFormat::GPTJ_2;
printf("\n---\nRetrying as GPT-J model: (ver %d)\nAttempting to Load...\n---\n", file_format);
lr = gpttype_load_model(inputs, file_format);
}
if (lr == ModelLoadResult::RETRY_LOAD)
{
file_format = FileFormat::GPTJ_3;
printf("\n---\nRetrying as GPT-J model: (ver %d)\nAttempting to Load...\n---\n", file_format);
lr = gpttype_load_model(inputs, file_format);
}
if(file_format==FileFormat::GPTJ_1)
{
//if we tried 1 first, then try 3 and lastly 2
//otherwise if we tried 3 first, then try 2
file_format = FileFormat::GPTJ_3;
printf("\n---\nRetrying as GPT-J model: (ver %d)\nAttempting to Load...\n---\n", file_format);
lr = gpttype_load_model(inputs, file_format);
}
//lastly try format 2
if (lr == ModelLoadResult::RETRY_LOAD)
{
file_format = FileFormat::GPTJ_2;
printf("\n---\nRetrying as GPT-J model: (ver %d)\nAttempting to Load...\n---\n", file_format);
lr = gpttype_load_model(inputs, file_format);
}
}
if (lr == ModelLoadResult::FAIL || lr == ModelLoadResult::RETRY_LOAD)
{
@@ -92,6 +99,19 @@ extern "C"
return true;
}
}
else if(file_format==FileFormat::RWKV_1)
{
printf("\n---\nIdentified as RWKV model: (ver %d)\nAttempting to Load...\n---\n", file_format);
ModelLoadResult lr = gpttype_load_model(inputs, file_format);
if (lr == ModelLoadResult::FAIL || lr == ModelLoadResult::RETRY_LOAD)
{
return false;
}
else
{
return true;
}
}
else
{
printf("\n---\nIdentified as LLAMA model: (ver %d)\nAttempting to Load...\n---\n", file_format);
@@ -102,7 +122,7 @@ extern "C"
generation_outputs generate(const generation_inputs inputs, generation_outputs &output)
{
if (file_format == FileFormat::GPTJ_1 || file_format == FileFormat::GPTJ_2 || file_format==FileFormat::GPTJ_3
|| file_format==FileFormat::GPT2_1 || file_format==FileFormat::GPT2_2 )
|| file_format==FileFormat::GPT2_1 || file_format==FileFormat::GPT2_2 || file_format==FileFormat::RWKV_1)
{
return gpttype_generate(inputs, output);
}