https://github.com/krshrimali/chatbot-rust-data-scrapping
part of code for generating chatbot - this includes data scraping for docs
https://github.com/krshrimali/chatbot-rust-data-scrapping
Last synced: 6 months ago
JSON representation
part of code for generating chatbot - this includes data scraping for docs
- Host: GitHub
- URL: https://github.com/krshrimali/chatbot-rust-data-scrapping
- Owner: krshrimali
- Created: 2024-02-15T17:13:55.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-03-02T16:31:07.000Z (over 1 year ago)
- Last Synced: 2025-03-12T08:26:10.468Z (7 months ago)
- Language: HTML
- Size: 30.3 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
## Goal
Input - documentation website (any documentation - pick PyTorch documentation website)
Output - A simple ChatBot server (try to build a UI out of iced-rs) that is able to infer based on the questions, and give meaningful results
## Method:
1. Be able to scrap the data from the docs website
2. Finetune a suitable model -> we'll just do it for minimum number of epochs
- Be able to figure out what model to use for fine-tuning
- What are the parameters for you to decide which model to use for fine-tuning?
3. Inference:
- How would you scale the inference model? --- Number of concurrent calls
- Performance!
- Fine-tuning we'll try to use simple plain PyTorch (python) for that, but also try torch-rs (rust bindings for PyTorch as an experimental thing)
- Definitely try inference from torch-rs / libtorch
Backup plan:
- Python server - Flask server that is called for a given question and the output is given to the user