Rohan's Notepad

My current goal

My initial work would be researching and creating an evaluation model for the FIM and also Researching how to efficiently stop AI hallucinations

My Research doc

https://docs.google.com/document/d/1LEzk-1YMWaU9O4_M6Jxog8uqWl43iH1s9CQ_-FQ3gSw/edit?tab=t.0#heading=h.n1php2j6f5pb

Resources

1. Syntax-Aware Fill-in-the-Middle (SAFIM): https://github.com/gonglinyuan/safim
2. Iterating over an answer and finding optimal code (might help in research about AI hallucination): https://www.linkedin.com/posts/stasbel_google-deepmind-just-dropped-a-bombshell-activity-7329140225190834176-M57l
3. TruthfulQA (OpenAI): https://arxiv.org/abs/2109.07958
4. A good ready-made evaluation tool: https://github.com/confident-ai/deepeval