I want to create a fine tuned completion model
which contains multiple “properties” inside the prompt.
reading the docs about data preparation
I used this stop sequence for each property \n
,
like
{"prompt":"Draft: some text\nText: some text", "completion":" this and that"}
going throw the examples in the docs I found the chat bot example
which uses different stop signs
{"prompt":"Summary: <summary of the interaction so far>\n\nSpecific information:<for example order details in natural language>\n\n###\n\nCustomer: <message1>\nAgent: <response1>\nCustomer: <message2>\nAgent:", "completion":" <response2>\n"}
{"prompt":"Summary: <summary of the interaction so far>\n\nSpecific information:<for example order details in natural language>\n\n###\n\nCustomer: <message1>\nAgent: <response1>\nCustomer: <message2>\nAgent: <response2>\nCustomer: <message3>\nAgent:", "completion":" <response3>\n"}
what is the reason to use this signs ?
<
\n\n
\n\n###\n\n
is there any further information about this formatting ?