In this video we see how to use pre-trained text embedding models from HuggingFace to embed movie reviews into a fixed size vector embedding. We can then perform clustering on these embeddings to group together reviews that have similar semantics! Using the labels for the reviews we can perform anomaly detection and see what reviews have been deemed positive reviews but are actually negative ones usually because the reviewer was being sarcastic!
Donations, Help Support this work!
https://www.buymeacoffee.com/lukeditria
The corresponding code is available here! (Section 16)
https://github.com/LukeDitria/pytorch...
Discord Server:
/ discord