try-750-seq-length-cerebros-attention-free-text

[david-thrower]

Kind of issue: Enhancement / R and D

Additional context: Enhancement to work in branch 131, merge candidate: 9e92a89a7b89c32e4c3244a97a1c323cff55bccc

Add any other context about the problem here. On the test set, we are getting binary_accuracy: 0.9397. We want to exceed 95%. Additionally, we would like to reduce the number of sub-trials required.

Suggested Labels (If you don't know, that's ok): kind/enhancement kind/accuracy kind/performance