small-mlm-glue-qnli-target-glue-mrpc

This model is a fine-tuned version of muhtasham/small-mlm-glue-qnli on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.9217
• Accuracy: 0.7770
• F1: 0.8455

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• training_steps: 5000

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.3905	4.35	500	0.7540	0.7892	0.8608
0.0675	8.7	1000	1.4012	0.7892	0.8608
0.0274	13.04	1500	1.5409	0.7794	0.8454
0.0189	17.39	2000	1.5464	0.7917	0.8609
0.0119	21.74	2500	1.7553	0.7794	0.8505
0.0179	26.09	3000	1.7660	0.7745	0.8492
0.0128	30.43	3500	1.9217	0.7770	0.8455

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.0+cu116
• Datasets 2.8.1.dev0
• Tokenizers 0.13.2