lingyit1108/ragtest-sakimilo

update knowledge in chroma db

resolved pdf reader issue

refined

swap to new embedding model and handle user 'i dont know' scenario

resolved dependency issues

tweak the vision_api prompt, create configuration files, minor tweak to main script

added winnie-the-pooh mode; and clean up question by user_id

fix qna_answer integer issue

switch default model to be gpt-4

added logical reasoning

added more questions

finishing up the app: UI, content understanding, questionaire, coaching

finished answer evaluation

corrected the streamlit query_params bug

to create RAGAs result with triad of metrics

finishing QnA and functions calling plus pydantic

cache chroma_db, fine-tuned-embeddings, etc.

reorganize files

added ux, vision_api, qna.txt

added streamlit-feedback deps

added feedback button

added fine-tuning notebook example

added fine-tuning example

added notebooks

added embedding artifacts

to implement rag in the streamlit chatbot app

added code to check time cost

added streamlit files under pages

added debugging code

corrected case-sensitive filename

point streamlit to streamlit_app

change hf color to the allowed color

added lfs debugging code

added environments info for reproducibility

added trulens implementation for evaluation

added all necessary pdfs

added HI Mock Exam Oct 2014 pdf; added gitattributes

added HI Chapter Summary pdf file

added compatible secrets management code

update readme to be compatible with huggingface

added github action workflows

tidy up single/double quotes

to fix minor bug

to fix the bug when NoneType occur at the end of prompt response

corrected code to generate openai completion delta content

to enable streaming feature in openai response

cap temperature to 2.0 as it is the maximum for openai gpt models

tidy up files and added requirements.txt

added all relevant assets for streamlit deployment

first commit