How do I get rid of ### Instruction, ### Answer, ### Explanation, Notes: in answers
Hi,
I have been playing around with a lot of models and your model is amazing @teknium
Whenever I prompt it in a language outside of english it will after the answer add these parameters:
"### Instruction"
"### Explanation"
"### Answer"
"### Answer & Explanation"
Notes:
Helpful Answer:
I have been using the model for RAG and whenever I prompt it in a non english language it adds new paragrafs like the example below, X and Y are companies I censored. I am asking about X:s acquisition of Y. After the answer has been sent the ### Instruction appears as below in english language.
My question: X har gjort en affär där de köpt Y. Berätta om den
Openhermes2 answer: X har köpt Y, vilket innebär att de nu äger och kontrollerar företaget. Detta är en affär som skedde år 2021.
"### Instruction:"
I want to know more about the acquisition of Y by X. Can you provide me with more information?
Here is my prompt template:
Open Hermes
messages = [
{
"role": "system",
"content": "You are a friendly swedish chatbot who always responds professionally to any questions. Always answer based on the context and tell the user if you are unable to find an answer. Always translate the answer to the Swedish language. Do not add ### Answer. Do not add ### Explanation. Do not add Notes:. Do not add ###Instruction to your messages.",
},
{"role": "user", "content": "Hi there!"},
{"role": "assistant", "content": "Nice to meet you!"},
{"role": "user", "content": "Can I ask a question?"}
]
Is there any way to get rid of this ?
Am I doing the prompt template wrong?
I dont think it is understanding the prompt format your giving it, but I need to see your code if your passing a json into it