Here is the space for a working Movie character “auto-chatbot”.
Movie Script finetuned LLama2-7f version , with Coqui.ai voices integrated (limited per hour).
Feel free to look at app.py to see how quantized (4bit via ggml) llama2 model is used and how custom chatbot is integrated (which includes some cleanup code as llama2-7f quantized version tends to produce some garbage like siempre, kwiek etc…)
Just a note: it is really slow on free 2 cpu one, so needed to use CPU upgrade with autoshutdown.