josueu/m2m100-T4-es-zap-data

m2m100-T4v1-es-zap-data

This model is a fine-tuned version of facebook/m2m100_418M on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.4157
• Bleu: 34.7034
• Ter: 51.7157
• Meteor: 0.5754
• Chrf: 58.5004

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Ter	Meteor	Chrf
6.2805	1.0	161	1.8069	13.5636	71.3235	0.4031	40.2762
2.5250	2.0	322	1.4623	21.4247	62.2549	0.4896	48.1914
1.6540	3.0	483	1.2635	22.4524	60.1716	0.4907	49.5555
0.8847	4.0	644	1.2306	28.2272	56.4951	0.5401	54.4320
0.5798	5.0	805	1.2738	25.9625	59.8039	0.5255	52.3034
0.3508	6.0	966	1.2854	32.8459	54.1667	0.5607	55.6667
0.2421	7.0	1127	1.3308	31.0852	54.0441	0.5463	55.4113
0.1745	8.0	1288	1.3297	29.9278	55.1471	0.5511	55.7676
0.1201	9.0	1449	1.3602	28.3774	56.25	0.5435	55.0030
0.0917	10.0	1610	1.3745	33.1321	51.8382	0.5732	57.6111
0.0567	11.0	1771	1.3614	33.4271	51.7157	0.5679	57.4082
0.0320	12.0	1932	1.3889	33.0171	52.2059	0.5655	57.8218
0.0279	13.0	2093	1.4141	33.4127	52.9412	0.5663	56.9120
0.0148	14.0	2254	1.4094	34.2420	52.4510	0.5722	57.5340
0.0132	15.0	2415	1.4157	34.7034	51.7157	0.5754	58.5004

Framework versions

• Transformers 5.4.0
• Pytorch 2.10.0+cu128
• Datasets 4.8.4
• Tokenizers 0.22.2