Model needs more training. Random baseline is 8 bpc. Good models achieve ~1.5-2.0 bpc.
Enter a context and click Predict to see model output.
| Source | Target | Strength |
|---|---|---|
| SPACE | '-' | |
| SPACE | '=' | |
| SPACE | '#' | |
| SPACE | '(' | |
| SPACE | 'f' | |
| SPACE | 'C' | |
| SPACE | 'b' | |
| SPACE | '<' |
Patterns extracted via compute_influence (Wx × Wy).
Many patterns weak due to undertrained model.