B-TIER demo
TensorBoard Embedding Projector — see what word vectors really associate
100k ft — What is this?
TensorBoard Embedding Projector — see what word vectors really associate
50k ft — Why care?
Uses the TensorBoard Embedding Projector to explore a word2vec model's neighborhoods: 'cold' pulls in Soviet and war, 'sexy' pulls in Groucho and wigs. A quick, visual way to show how embeddings absorb history and cultural bias from training text.
10k ft — Key takeaways
• Embeddings encode historical and cultural associations, not just definitions — useful for teaching bias
• Load any embedding model into the Projector to inspect nearest neighbors interactively
5k ft — Tools & techniques
TensorBoard, TensorBoard Projector
nearest-neighbor search, word embeddings
1k ft — Deep-dive / try it
Tensorboard embedding projector - repost
Transcript excerpt
Ready to see how AI thinks about the world? Take a look at cold. We'll see some concepts like hot and spring that we associate with cold. But this AI has also learned history, and you'll see concepts like Soviet and war are also related to cold. I'm using the TensorBoard embedding projector to be able to see these embeddings or vector representations of words, in this case a word-to-vec model. Let's look at sexy. You'll see some of the concepts that you're probably associated with, such as beard and curvy. But look at this AI. It also associates groucho and wigs with sexy.