a6ace61febf24ad62e27a3dd33dbfa4a

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

• Loss: 2.1269
• Data Size: 1.0
• Epoch Runtime: 44.3674
• Bleu: 5.0572

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	20.8328	0	3.6571	0.0136
No log	1	91	20.5809	0.0078	4.7164	0.0176
No log	2	182	19.0958	0.0156	5.7024	0.0111
No log	3	273	18.4367	0.0312	7.8785	0.0133
No log	4	364	17.6781	0.0625	10.5731	0.0110
No log	5	455	14.6516	0.125	13.8143	0.0139
No log	6	546	6.4806	0.25	18.6018	0.0296
1.5695	7	637	3.2936	0.5	28.5055	0.2850
3.8773	8.0	728	2.4982	1.0	45.8832	1.3749
3.0837	9.0	819	2.3093	1.0	43.1954	3.6330
2.789	10.0	910	2.2261	1.0	43.5214	4.0019
2.5661	11.0	1001	2.1749	1.0	46.5535	4.2296
2.4472	12.0	1092	2.1430	1.0	44.4746	4.3623
2.3362	13.0	1183	2.1247	1.0	43.9003	4.4396
2.2216	14.0	1274	2.1092	1.0	45.1125	4.5845
2.1228	15.0	1365	2.1020	1.0	44.5265	4.5838
2.0104	16.0	1456	2.0972	1.0	44.4803	4.6869
1.9408	17.0	1547	2.0954	1.0	45.7054	4.7448
1.8473	18.0	1638	2.0965	1.0	44.7305	4.8875
1.7498	19.0	1729	2.1176	1.0	43.8897	4.9420
1.6988	20.0	1820	2.1216	1.0	46.3713	5.0510
1.6431	21.0	1911	2.1269	1.0	44.3674	5.0572

Framework versions

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