How does this work?

Trained on the complete works of Shakespeare, this character-based model generates text in the style of Shakespeare.

Training Steps

Generate text from input
The final result is the model scaled up to 36 million parameters. The output is almost entirely made up of real words.
Train the model by character
The final result is the model scaled up to 36 million parameters. The output is almost entirely made up of real words.
Build the model's attention
The final result is the model scaled up to 36 million parameters. The output is almost entirely made up of real words.
Optimize the model
The final result is the model scaled up to 36 million parameters. The output is almost entirely made up of real words.
Final Result - Scaling up
The final result is the model scaled up to 36 million parameters. The output is almost entirely made up of real words.
Bonus: Jane Austen
The final result is the model scaled up to 36 million parameters. The output is almost entirely made up of real words.