b075eb4a9e29a1e6571d7047473604c5

This model is a fine-tuned version of google/mt5-base on the Helsinki-NLP/opus_books [fr-nl] dataset. It achieves the following results on the evaluation set:

• Loss: 1.6085
• Data Size: 1.0
• Epoch Runtime: 209.5152
• Bleu: 9.7936

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	15.4205	0	17.2439	0.0154
No log	1	1000	13.9017	0.0078	19.7549	0.0194
No log	2	2000	11.4029	0.0156	20.5286	0.0212
No log	3	3000	10.3222	0.0312	24.3867	0.0179
0.4118	4	4000	5.8715	0.0625	31.3563	0.0230
4.4572	5	5000	2.9395	0.125	42.7468	2.0196
0.2117	6	6000	2.5269	0.25	66.7118	3.8018
0.2683	7	7000	2.3041	0.5	116.0491	4.8542
2.6311	8.0	8000	2.1122	1.0	210.4427	5.7843
2.4335	9.0	9000	2.0011	1.0	207.4353	6.4811
2.2967	10.0	10000	1.9233	1.0	207.8650	6.9229
2.2179	11.0	11000	1.8646	1.0	207.1371	7.3574
2.1122	12.0	12000	1.8194	1.0	206.8493	7.7190
2.0336	13.0	13000	1.7927	1.0	207.7914	7.9566
1.9738	14.0	14000	1.7589	1.0	210.8171	8.2123
1.8977	15.0	15000	1.7363	1.0	209.6291	8.3925
1.8554	16.0	16000	1.7115	1.0	210.7359	8.5549
1.7706	17.0	17000	1.6998	1.0	211.4426	8.7277
1.7873	18.0	18000	1.6766	1.0	211.3672	8.8573
1.7265	19.0	19000	1.6663	1.0	209.6792	8.9481
1.672	20.0	20000	1.6533	1.0	210.9265	9.1015
1.6525	21.0	21000	1.6385	1.0	210.8974	9.1918
1.6174	22.0	22000	1.6344	1.0	211.5922	9.3196
1.5776	23.0	23000	1.6296	1.0	212.8005	9.3744
1.5264	24.0	24000	1.6211	1.0	209.8647	9.4105
1.5552	25.0	25000	1.6154	1.0	210.6224	9.5693
1.4744	26.0	26000	1.6045	1.0	209.0745	9.5901
1.4567	27.0	27000	1.6080	1.0	209.5370	9.6895
1.4102	28.0	28000	1.6087	1.0	209.9979	9.6693
1.4109	29.0	29000	1.6057	1.0	209.9585	9.7511
1.3673	30.0	30000	1.6085	1.0	209.5152	9.7936

Framework versions

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