m2m100_418M-ft-cy-to-en

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

• Loss: 0.8103
• Bleu: 38.1416
• Gen Len: 30.6116

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 6000
• training_steps: 30000

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
2.1452	0.0166	2000	1.8458	16.5803	33.8627
1.6103	0.0332	4000	1.3438	24.8191	32.3474
1.3416	0.0499	6000	1.1426	30.3046	31.0528
1.2254	0.0665	8000	1.0307	28.6069	33.3047
1.1184	0.0831	10000	0.9745	34.3431	30.2578
1.0831	0.0997	12000	0.9310	33.0402	31.6276
1.0426	0.1164	14000	0.8990	36.4181	29.894
1.018	0.1330	16000	0.8783	37.2119	29.5248
0.9957	0.1496	18000	0.8601	37.9897	30.2633
0.9796	0.1662	20000	0.8485	37.5805	31.3311
0.9687	0.1829	22000	0.8378	39.4133	29.8813
0.9568	0.1995	24000	0.8270	38.5896	30.7235
0.9377	0.2161	26000	0.8169	38.7364	30.7135
0.942	0.2327	28000	0.8125	37.8857	30.7208
0.9427	0.2494	30000	0.8103	38.1416	30.6116

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0