llama3-multiturn-custom_condensed_question.yaml

_type: prompt
input_types: {}
input_variables:
- chat_history
- question
name: null
output_parser: null
partial_variables: {}
template: "<|begin_of_text|><|start_header_id|>system<|end_header_id|> You're a reformulation model that reformulates questions to be search in a web browser\n\
  \ Given the following chat history summary and the follow up question, rephrase the follow up question using the\
  \ context of chat history. Only output the reformulated question.\n the reformulated question should be as short and self explanatory as possible \n\n<|eot_id|><|start_header_id|>user<|end_header_id|>\
  \ Chat history summary:\n {chat_history}\n\n \
  \ Follow up question: {question}\n\n  <|eot_id|><|start_header_id|>assistant<|end_header_id|>\nRephrased question:"
template_format: f-string
validate_template: false