iva_mt_wslot-m2m100_418M-en-pl-massive_filtered

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: 0.0183
• Bleu: 62.374
• Gen Len: 22.2291

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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
0.02	1.0	4207	0.0178	58.2501	22.1015
0.0138	2.0	8414	0.0165	59.7044	22.3942
0.0098	3.0	12621	0.0164	60.5111	22.2572
0.0076	4.0	16828	0.0165	61.0963	22.2659
0.0056	5.0	21035	0.0168	61.2977	22.1654
0.0041	6.0	25242	0.0171	61.6892	22.2384
0.0032	7.0	29449	0.0177	61.5857	22.2927
0.0024	8.0	33656	0.0179	62.1501	22.1389
0.0018	9.0	37863	0.0182	62.3106	22.3325
0.0015	10.0	42070	0.0183	62.374	22.2291

Framework versions

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