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

merge new rope param nonsense

Browse source
This commit is contained in:
Concedo 2023-09-30 11:33:30 +08:00
commit b84e210f0d
38 changed files with 2811 additions and 764 deletions
Download patch file Download diff file Expand all files Collapse all files

4
gpttype_adapter.cpp
View file

@ -832,7 +832,7 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
n_vocab = llama_n_vocab(llama_ctx_v4);
//determine mem per token
const std::vector<int> tmp = {1, 2, 3, 4};
std::vector<int> tmp = {1, 2, 3, 4};
auto er = llama_eval(llama_ctx_v4, tmp.data(), tmp.size(), 0, params.n_threads);
if(er!=0)
{
@ -1465,7 +1465,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs, generation_o
::utreplace(tmp, "\n", "\\n");
outstr += tmp;
outstr += "\n\n[Debug: Context Size = " + std::to_string(current_context_tokens.size()) + "]\n";
outstr += "\n\n[Debug: n_past="+std::to_string(n_past)+" Context Size = " + std::to_string(current_context_tokens.size()) + "]\n";
tmp = "";
for (auto id : current_context_tokens)
{