diff --git a/expose.h b/expose.h index 27bc391cd..4a509be5c 100644 --- a/expose.h +++ b/expose.h @@ -4,7 +4,7 @@ const int tensor_split_max = 16; const int images_max = 8; const int audio_max = 4; -const int logprobs_max = 5; +const int logprobs_max = 10; const int overridekv_max = 4; // match kobold's sampler list and order diff --git a/gpttype_adapter.cpp b/gpttype_adapter.cpp index 81f22e08f..54c667fdc 100644 --- a/gpttype_adapter.cpp +++ b/gpttype_adapter.cpp @@ -4732,7 +4732,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs) std::string topstr = toppick.selected_token; ::utreplace(topstr, "\n", "\\n"); printf("(%s <%d> %.2f%%)", RemoveBell(topstr).c_str(), toppick.selected_tokenid, toppick.selected_probability*100); - int maxtoshow = (toppick.tokenid.size()>4?4:toppick.tokenid.size()); + int maxtoshow = (toppick.tokenid.size()>4?4:toppick.tokenid.size()); //hardcode limit even if we have more logprobs_max for (int i=0;i