- Q, question
- one
- group by number, for example 3
- only for calling api
- group by category
- TODO
- all
- manul input question
- from CM:REPL to here
- P, paper
- one
- group by selection
- all
- PT, template:
- one per question
- batch of question
- M, model
- GPT
- openLLM
- TODO
- langchain
- other model
- CM, chat mode
- REPL when no question
- auto run
- PC, Paper content
- all content
- map reduce
- embedding/semantic search
- refine mode
- PQ, Prompt question method
- ask Q one by one
- ask some Q in one request
- paper
- question_id
- question
- question_type
- human_answer
- AI_reply
- AI_answer
- correct_answer?
- AI_NA
- Comment
- explain
- correct_explain?
- sentences
- correct_sentences?
- confidence
- completion_tokens
- prompt_tokens
- total_tokens
- seconds
- model
- chat_mode
- date
- md5
- #batches