51cdf3937d77d63d1909727088eba455

This model is a fine-tuned version of google-bert/bert-base-multilingual-uncased on the nyu-mll/glue [stsb] dataset. It achieves the following results on the evaluation set:

• Loss: 0.5816
• Data Size: 1.0
• Epoch Runtime: 11.1985
• Mse: 0.5819
• Mae: 0.5836
• R2: 0.7397

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	Mse	Mae	R2
No log	0	0	8.3808	0	1.1760	8.3821	2.4785	-2.7496
No log	1	179	5.9311	0.0078	1.5453	5.9323	2.0179	-1.6537
No log	2	358	4.9055	0.0156	1.5576	4.9066	1.8298	-1.1949
No log	3	537	2.6245	0.0312	1.9641	2.6253	1.3405	-0.1744
No log	4	716	3.4282	0.0625	2.3898	3.4288	1.4940	-0.5338
No log	5	895	1.0045	0.125	2.9029	1.0047	0.7992	0.5506
0.1552	6	1074	1.1201	0.25	4.1852	1.1203	0.7882	0.4988
0.8744	7	1253	0.7305	0.5	6.7032	0.7308	0.6780	0.6731
0.611	8.0	1432	0.6129	1.0	11.3890	0.6132	0.6138	0.7257
0.399	9.0	1611	0.6787	1.0	10.9921	0.6790	0.6117	0.6963
0.2543	10.0	1790	0.5821	1.0	11.2857	0.5825	0.5886	0.7394
0.2029	11.0	1969	0.5743	1.0	10.8639	0.5746	0.5853	0.7429
0.1701	12.0	2148	0.5462	1.0	10.4000	0.5466	0.5589	0.7555
0.1353	13.0	2327	0.5787	1.0	10.5432	0.5789	0.5813	0.7410
0.1299	14.0	2506	0.5512	1.0	10.5992	0.5514	0.5645	0.7533
0.1203	15.0	2685	0.6181	1.0	10.6529	0.6183	0.6088	0.7234
0.0924	16.0	2864	0.5816	1.0	11.1985	0.5819	0.5836	0.7397

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1