Inspiration
The advantage of having a quick time to market and the benefits we can reap
What it does
It combines the customer review and the personas we created and provides opinions on design ideas to product developers
How we built it
We used the databricks ecosystem to harness the LLama , Langchain and chroma db to convert the review contents into chunk of documents, embed it using langchain, store it into chroma db. After this when user makes a query, we used langchain pipelines, fed the model the context that is retrieved using RAG, the personas that the user selected and gave back response
Challenges we ran into
Hallucination, unwanted response, GPU size , UI/ UX
Accomplishments that we're proud of
We were able to give the users the ability to get opinion of people with different mindset using persona. This gives the ability to consider trade offs based in user opinions of various demographics.
What we learned
LLM are powerful tool , it has challenges. But databricks has rich ecosystem to help developers to create powerful applications with limited resources
What's next for LLM's in Automotive
Scale it with more data and personas Include sentiment meter Consider fine tuning over RAG Improve pre processsing Build a UI and float the tool to our vehicle design engineers
Built With
- chroma
- databricks
- huggingface
- langchain
- llama
Log in or sign up for Devpost to join the conversation.