hello? I am a professor of internal medicine at Chungbuk National University College of Medicine in Korea.
I am teaching allergy to students, and I have a question.
Rather than a medical examination score, I would like to create a question and answer + explanation bot or GPTs that students can use for learning.
I have a list of exam questions-correct answers-commentary for medical school students, which can be converted into a format similar to 'gpt_chain_of_thougts_with_context' in prompt_template.py of promptbase - MMLU.
I want to create a Q&A bot using MedPrompt papers.
(1) Should I transform the questions I have to match all prompt _names in the prompt_template.py file?
(2) As shown in the paper, I would like to maximize performance by combining the dynamic few shots, senf-generated chain of thoug, and majority vote ensenbling methods. Can I refer to a file combining the above three methods on the github page?
hello? I am a professor of internal medicine at Chungbuk National University College of Medicine in Korea.
I am teaching allergy to students, and I have a question.
Rather than a medical examination score, I would like to create a question and answer + explanation bot or GPTs that students can use for learning.
I have a list of exam questions-correct answers-commentary for medical school students, which can be converted into a format similar to 'gpt_chain_of_thougts_with_context' in prompt_template.py of promptbase - MMLU.
I want to create a Q&A bot using MedPrompt papers.
(1) Should I transform the questions I have to match all prompt _names in the prompt_template.py file?
(2) As shown in the paper, I would like to maximize performance by combining the dynamic few shots, senf-generated chain of thoug, and majority vote ensenbling methods. Can I refer to a file combining the above three methods on the github page?