Keeps repeating same reply.
#2
by
Bharatvid
- opened
Its a good model, but it works only once, after that, it just keep repeating the same reply. I am using Q4_K_M with llama.cpp. I also tried with Koboldcpp and Text Generation Web UI, but same problem.
If I have to translate anything, I have to put the texts in Settings > System Message in llama.cpp. Thats work around. Or I have to open the new tab and do it again.