t5-small-finetuned_entailment_inference

This model is a fine-tuned version of t5-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.0249

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.2885	1.0	522	1.2005
1.2209	2.0	1044	1.1594
1.1871	3.0	1566	1.1263
1.1455	4.0	2088	1.1098
1.1124	5.0	2610	1.0949
1.0758	6.0	3132	1.0825
1.0485	7.0	3654	1.0707
1.0205	8.0	4176	1.0606
0.9913	9.0	4698	1.0523
1.0099	10.0	5220	1.0463
0.97	11.0	5742	1.0395
0.9699	12.0	6264	1.0370
0.9531	13.0	6786	1.0337
0.9449	14.0	7308	1.0312
0.9354	15.0	7830	1.0274
0.9342	16.0	8352	1.0266
0.9188	17.0	8874	1.0262
0.9219	18.0	9396	1.0251
0.9044	19.0	9918	1.0252
0.9223	20.0	10440	1.0249

Framework versions

• Transformers 4.24.0
• Pytorch 1.12.1+cu113
• Datasets 2.7.1
• Tokenizers 0.13.2