While this will get you a selfhosted LLM it is not possible to feed data to them like this. As far as I know there are a 2 possibilities:
-
Take an existing model and use the literature data to fine tune the model. The success of this will depend on how much “a lot” means when it comes to the literature
-
Create a model yourself using only your literature data
Both approaches will require some yrogramming knowledge and understanding of how a llm works. Additionally it will require a preparation of the unstructured literature data to a kind of structured data that can be used to train or fine tune the model.
Im just a CS student so not an expert in this regard ;)
Using “AI” has been beneficial for example to generate image descriptions automatically, which were then used as alternative text on a website. This increased accessibility AND users were able to use full text search on these descriptions to find images faster. Same goes for stuff like classification of images, video and audio. I know of some applications in agriculture where object detection and classification etc. is used to optimize the usage of fertilizer and pesticides reducing costs and reducing environmental impact they cause. There are ofcourse many more examples like these but the point should be clear.