vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-10 17:14:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

tidy up and refactor code to support old flags

Browse source
This commit is contained in:
Concedo 2024-05-10 16:50:53 +08:00
parent eccc2ddca2
commit dbe72b959e
4 changed files with 88 additions and 173 deletions
Download patch file Download diff file Expand all files Collapse all files

4
gpttype_adapter.cpp
View file

@ -92,7 +92,6 @@ static int current_llava_identifier = LLAVA_TOKEN_IDENTIFIER_A;
static gpt_params * kcpp_params = nullptr;
static int max_context_limit_at_load = 0;
static int n_past = 0;
static bool useSmartContext = false;
static bool useContextShift = false;
static int debugmode = 0; //-1 = hide all, 0 = normal, 1 = showall
static std::string modelname;
@ -787,7 +786,6 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
}
kcpp_params->flash_attn = inputs.flash_attention;
modelname = kcpp_params->model = inputs.model_filename;
useSmartContext = inputs.use_smartcontext;
useContextShift = inputs.use_contextshift;
debugmode = inputs.debugmode;
@ -1941,7 +1939,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
}
else
{
bool triggersc = useSmartContext;
bool triggersc = useContextShift;
if(useContextShift && (file_format == FileFormat::GGUF_GENERIC))
{
PurgeMissingTokens(llama_ctx_v4, current_context_tokens, embd_inp, inputs.max_length, nctx);