mASI use of DNN and Language Model APIs (Walk-Through Excerpt)


The following is part of the code-level walk-through of mASI technology we’re preparing to release.

Previously we have walked through how the code over the simple case works, including mediation processing. In figure 17, There are a couple of calls to methods on the ‘TheContextDB’ object. This object essentially is part of the context engine and wraps the context graph database. The last part of this block creates the knowledge graph that goes into the mediation queue. These calls use DNN (Deep Neural network) based Machine Learning APIs similar to GPT-3. What we are going to do is a walkthrough of how this works using GPT-3. Meaning to do the test here, we swapped out the GPT-3 as the first API and an API like Grammarly as the second API. The approach is a different thing, just using the API straight up so we will talk through the execution and show you how even using GPT-3 in place of this service produces similar results when used with this methodology.

Continue reading “mASI use of DNN and Language Model APIs (Walk-Through Excerpt)”