From: Improving deep neural network design with new text data representations
Experiment | Partition | Num. of instances | % of data |
---|---|---|---|
Embedding | Full | 100,000 | 100 |
Training | 72,000 | 72 | |
Validation | 12,960 | 18 | |
Test | 1296 | 10 | |
Padding | Full | 1600,000 | 100 |
Training | 1152,000 | 72 | |
Validation | 207,360 | 18 | |
Test | 20,736 | 10 |