README.md

metadata

language: en
license: mit
library_name: pytorch

Plainly Optimized Network

Dataset: BIGBENCH

Trainer Hyperparameters:

• lr = 5e-05
• per_device_batch_size = 1
• gradient_accumulation_steps = 4
• weight_decay = 1e-09
• seed = 42

eval_loss	eval_accuracy	epoch
66.323	0.063	1.0
59.935	0.055	2.0
60.344	0.056	3.0
58.559	0.054	4.0
56.373	0.051	5.0
58.011	0.053	6.0