t5-small-finetuned-de-en-final

This model is a fine-tuned version of t5-small on the wmt14 dataset. It achieves the following results on the evaluation set:

• Loss: 2.3285
• Bleu: 9.8394
• Gen Len: 17.325

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: 0.0002
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	188	2.3867	9.7928	17.2581
No log	2.0	376	2.3942	9.7222	17.4186
0.7948	3.0	564	2.3909	9.6495	17.3513
0.7948	4.0	752	2.3496	9.7376	17.3417
0.7948	5.0	940	2.3285	9.8394	17.325

Framework versions

• Transformers 4.12.5
• Pytorch 1.10.0+cu111
• Datasets 1.16.1
• Tokenizers 0.10.3