iva_mt_wslot-m2m100_418M-en-es-massive_unfiltered

This model is a fine-tuned version of facebook/m2m100_418M on the iva_mt_wslot-exp dataset. It achieves the following results on the evaluation set:

• Loss: 0.0114
• Bleu: 67.6426
• Gen Len: 18.9134

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 7
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
0.0129	1.0	2879	0.0118	65.4383	18.8697
0.009	2.0	5758	0.0109	66.6878	18.9331
0.0066	3.0	8637	0.0107	66.6143	18.8687
0.0049	4.0	11516	0.0108	66.9832	18.8067
0.0037	5.0	14395	0.0109	67.452	18.8598
0.0028	6.0	17274	0.0112	67.4281	18.9213
0.0023	7.0	20153	0.0114	67.6426	18.9134

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3