This is copy of the Flask server and code set up for Project SleepeR a AI feature for nodenogg.in to intergrate with small data sets of reading lists and surface recommended content to students/ researchers using nodenogg.in via human understandable AI, with the option to dial up the AI as you felt happy about. This PoC using LLMs that are on Hugging face but the models that some have been trained on are within Blackboxes and would thus not fit the remit of a release version but the PoC allowed us to demonstrate the potential of small data sets and small understandable AI.
The orignal code for this is hosted at gitlab.adamprocter.co.uk and a Google Collob to support some of the documentation was also created https://colab.research.google.com/drive/1xQPbrWMFIKz76yzaCdrTx1N6OnHByncR?usp=sharing