18-24-finetuned-eng-to-para

This model is a fine-tuned version of Tri1/12-18-finetuned-eng-to-para on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3257
• Bleu: 17.0341
• Gen Len: 24.16

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: 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: 6

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
0.1696	1.0	6250	0.3216	18.5922	24.048
0.1554	2.0	12500	0.3225	17.7026	23.992
0.1474	3.0	18750	0.3242	17.2459	24.176
0.1387	4.0	25000	0.3243	17.3668	23.856
0.1314	5.0	31250	0.3247	17.4414	24.416
0.1277	6.0	37500	0.3257	17.0341	24.16

Framework versions

• Transformers 4.30.1
• Pytorch 2.0.0
• Datasets 2.1.0
• Tokenizers 0.13.3