mt5-base_2_3

This model is a fine-tuned version of obokkkk/mt5-base_2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.1465
• Bleu: 9.5474
• Gen Len: 17.854

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: 0.0001
• 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
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	175	1.1739	9.0271	17.8543
No log	2.0	350	1.1660	9.1398	17.8468
1.3653	3.0	525	1.1585	9.251	17.8656
1.3653	4.0	700	1.1538	9.3176	17.8476
1.3653	5.0	875	1.1518	9.3529	17.8608
1.2985	6.0	1050	1.1505	9.4818	17.8552
1.2985	7.0	1225	1.1475	9.499	17.8575
1.2985	8.0	1400	1.1471	9.5511	17.871
1.2632	9.0	1575	1.1459	9.5315	17.8547
1.2632	10.0	1750	1.1465	9.5474	17.854

Framework versions

• Transformers 4.18.0
• Pytorch 1.11.0+cu113
• Datasets 2.1.0
• Tokenizers 0.12.1