BertViz: Visualize Attention in NLP Models (BERT, GPT2, BART, etc.)
BertViz is an interactive tool for visualizing attention in Transformer language models such as BERT, GPT2, or T5. It can be run inside a Jupyter or Colab notebook through a simple Python API that supports most Huggingface models. BertViz extends the Tensor2Tensor visualization tool by Llion Jones, providing multiple views that each offer a unique lens into the attention mechanism. The head view visualizes attention for one or more attention heads in the same layer. It is based on the...
Siafu simulates individual agents and their context, from home to city-wide scenarios. As a developer, you use the API to write your simulation for the purposes of data-set generation, test or visualization, optionally hooking it to your own application.
The purpose of SNX is to have a modular series of components that report back to a central repository and 'AI' for pattern recognition and anomoly detection. This is to help with IDS systems, and to move to the next generation of security.