c44d84929fae3e8cd47adb97b772d4f6

This model is a fine-tuned version of google/mt5-xl on the Helsinki-NLP/opus_books [fr-no] dataset. It achieves the following results on the evaluation set:

• Loss: 2.0325
• Data Size: 1.0
• Epoch Runtime: 54.2113
• Bleu: 7.1423

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: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Bleu
No log	0	0	7.7037	0	3.3916	0.0152
No log	1	86	5.9013	0.0078	3.8969	0.0374
No log	2	172	4.4377	0.0156	11.3199	0.1792
No log	3	258	3.8921	0.0312	19.7112	0.3155
No log	4	344	3.4667	0.0625	26.3629	0.5115
0.2713	5	430	2.7973	0.125	22.0126	0.7753
0.8003	6	516	2.2375	0.25	26.1557	4.8100
0.9236	7	602	2.0360	0.5	39.4281	5.3175
1.3034	8.0	688	1.9308	1.0	60.7768	6.2436
1.9059	9.0	774	1.9122	1.0	49.9668	6.6599
1.7108	10.0	860	1.9274	1.0	56.1671	6.9248
1.5129	11.0	946	1.9303	1.0	52.1314	7.0669
1.315	12.0	1032	1.9777	1.0	50.1375	7.1446
1.1926	13.0	1118	2.0325	1.0	54.2113	7.1423

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.1