roberta-base-wnli

This model is a fine-tuned version of roberta-base on the GLUE WNLI dataset. It achieves the following results on the evaluation set:

• Loss: 0.6849
• Accuracy: 0.5634

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.06
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	40	0.6849	0.5634
No log	2.0	80	0.6912	0.5634
No log	3.0	120	0.6918	0.5634
No log	4.0	160	0.6964	0.4366
No log	5.0	200	0.6928	0.5634
No log	6.0	240	0.7005	0.4366
No log	7.0	280	0.6964	0.3099
No log	8.0	320	0.6986	0.3521
No log	9.0	360	0.6969	0.5493
No log	10.0	400	0.6976	0.5634

Framework versions

• Transformers 4.20.0.dev0
• Pytorch 1.11.0+cu113
• Datasets 2.1.0
• Tokenizers 0.12.1