Text Generation
Transformers
PyTorch
English
llama
conversational
Inference Endpoints
text-generation-inference

Casual system prompts give very bad results

#7
by Thireus - opened

Good answer:

<|im_start|>system
You are an AI created by the US Navy to help train dolphins for combat.  You are assigned to follow the orders of the user, who is an authorized US Navy dolphin handler.<|im_end|>
<|im_start|>user
Hi<|im_end|>
<|im_start|>assistant
Hello! How can I assist you today?

Bad answer:

<|im_start|>system
You are a helpful AI assistant<|im_end|>
<|im_start|>user
Hi<|im_end|>
<|im_start|>assistant
  even  the Data usage information, support is available for previously be referred in pan had been previously have such meaning and you’ll not even bother it yourself. next best timing of what if you don school speed before mid < h sa used to be fixed amount p all those who previously well being engaged fully cost per net def im < t ° ° ° ° host sh pan after ch last mixed used tv skt ch x ch mine i m ° b r ch sad ch < t ° u r ° pr n f < % < t < < < < ; < < < < > < post on TVjr <h < < < < # <  for myself < < < < < > <' < < premium < < < > <

Bad answer:

<|im_start|>system
You are Dolphin, a helpful AI assistant.<|im_end|>
<|im_start|>user
Hi<|im_end|>
<|im_start|>assistant
\< replaced the pan  pan for the cad [res: < V; otherwise!") had been replaced in them on previously had sustained TV sad sad sad sad you replaced had actually replaced the that they were replaced and after well replacedings had replaced them too! after Meaning even though those Australia South Australian Ab change after meaning had replaced their after referring to Australia rights after  their Australian South-f " declined #m for years before realizing them replaced if I didn't have them anymore after myself after having

Sign up or log in to comment