Recursively Neural

Funny? Funny, how? In what way? Why is that funny? Is this some kind of a joke?

That which we cannot joke about, we must pass over in silence.

1. Transformers

• Encoder-decoder framework
• Attention mechanisms
• Transfer learning

2. Text Classification

• Datasets
• Tokenizers for Sub-Words, Words, Sentences, Paragraphs, Documents
• Text Analytics

3. Transformer Anatomy

• Getting the embedded vectors or the hidden state or context
• Iterative generation, brainstorming, riffing without the meeting hogs, clowns, attention whores

4. Calling Out Multijargon Stupid-Named Entity misRecognition

• Double entrendre
• Ambiguity
• Puns, malapropisms
• Finer points
• Distinction
• No, you don't understand ... you think you do ... but you don't EVEN understand that you don't understand.

5. Joke Generation

6. Bad Summarization and Missing The Point To Make A Point

7. Question Misanswering and Dodging

8. Making Transformers Funny In Production

9. Dealing With Few To No Labels

10. Training Transformers From Scratch

11. Future Directions