Week 6
Finished
Developed JSON reader for llama-index
- Improved upon the foundation of Rohan’s JSON parser
- Leave out the irrelevant information and formats
- Keep core FrameNet data skeleton
- Applied JSON Reader in llama-index
- Reference: LlamaIndex JSON reader guide
Developed multi-conversation function
- Multi-round conversation
- Store memory of previous conversation
More Prompting Examples
Please refer to Red Hen Meeting - Jul 03
Other Messages From The Weekly Meeting
Looking for a systemetic way to construct the prompt generation algorithm
- Frame selection / categorization
- Frame-frame relationship (e.g. inheritance, usage, …)
- Lexical Units - specify phrase structure and lexical units
- Full Text Annotation - specify frames, it can be a natural few shots prompt
- Observation:
- How do choice of worlds change the results
- How do Chain of thoughts increase performance and explainability
- How do Zero shot, one shot, few shot change the results
To-do List For Next Week
- Continue reading books and CoT papers
- Work on prompt generating algorithm