t5-base-vanilla-top_v2

This model is a fine-tuned version of google/mt5-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0349

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: 0.001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 64
• total_train_batch_size: 512
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 3000

Training results

Training Loss	Epoch	Step	Validation Loss
0.9322	0.82	200	0.0776
0.0758	1.65	400	0.0476
0.0732	2.47	600	0.0762
0.0527	3.29	800	0.0416
0.0448	4.12	1000	0.0449
0.0401	4.94	1200	0.0368
0.0312	5.76	1400	0.0351
0.0268	6.58	1600	0.0353
0.0239	7.41	1800	0.0351
0.0219	8.23	2000	0.0353
0.0202	9.05	2200	0.0346
0.0184	9.88	2400	0.0345
0.017	10.7	2600	0.0352
0.0163	11.52	2800	0.0351
0.0154	12.35	3000	0.0349

Framework versions

• Transformers 4.24.0
• Pytorch 1.13.0+cu117
• Datasets 2.7.0
• Tokenizers 0.13.2