when we were trying out the markitdown files on research papers it created a lot of noise from the papers so we tried searching for an alternative which was docling - which is a massive ml model which give noiseless readme.(1/n)
when we were trying out the markitdown files on research papers it created a lot of noise from the papers so we tried searching for an alternative which was docling - which is a massive ml model which give noiseless readme.(1/n)
but docling only works on small range of file types and docling works better wihth gpu so we tried combining both markdown and docling as two different modes one for research papers and each general and research modes have varying level of optimization(2/n)
@Radicalpredator@jaidev_narayan@ilovepdf_com The purpose of this account and the projects is to build and deploy for public in public and also take the help of people in building a better product overall. https://t.co/QIabEOVBsc
this is the current product we have.
Video of the product in use coming soon (4/n)
@Radicalpredator@jaidev_narayan We thought we can add optimization of token on this and other features which would be very beneficial while uploading a context to a llm. so we built more sophisticated (obviously vibe coded website) which also had a optimization feature in it. (2/n)
@Radicalpredator@jaidev_narayan But our main idea is to deploy this bad boy and make it free for all. We are somewhat fans of open source and @ilovepdf_com . so we are trying to build something similar but we have a lots of issues to solve before we deploy the first version.(3/n)
So one day me @Radicalpredator and @jaidev_narayan found https://t.co/fcIqMuMCol and thought we could build something from it.
the initial prototype was built and hosted in streamlit.
https://t.co/pmB1gLldQp
but we felt there is a lot of improvements we can do on this.(1/n)
So one day me @Radicalpredator and @jaidev_narayan found https://t.co/fcIqMuMCol and thought we could build something from it.
the initial prototype was built and hosted in streamlit.
https://t.co/pmB1gLldQp
but we felt there is a lot of improvements we can do on this.(1/n)
@Radicalpredator@jaidev_narayan@ilovepdf_com The purpose of this account and the projects is to build and deploy for public in public and also take the help of people in building a better product overall. https://t.co/QIabEOVBsc
this is the current product we have.
Video of the product in use coming soon (4/n)