f0c7e0f7b6218a1a5515caa149330136

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

• Loss: 0.6721
• Data Size: 1.0
• Epoch Runtime: 7.1234
• Mse: 0.6725
• Mae: 0.6130
• R2: 0.6992

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	7.6286	0	0.9834	7.6299	2.3284	-2.4131
No log	1	179	5.5486	0.0078	1.2799	5.5498	1.9497	-1.4826
No log	2	358	3.0790	0.0156	1.5146	3.0800	1.4825	-0.3778
No log	3	537	2.4242	0.0312	1.6751	2.4249	1.3003	-0.0847
No log	4	716	2.1537	0.0625	1.9012	2.1543	1.2054	0.0363
No log	5	895	1.0065	0.125	2.3028	1.0068	0.7942	0.5496
0.1232	6	1074	0.9907	0.25	3.0432	0.9913	0.7701	0.5566
0.754	7	1253	0.7712	0.5	4.2346	0.7716	0.6790	0.6548
0.6054	8.0	1432	0.6719	1.0	7.2157	0.6723	0.6306	0.6993
0.3795	9.0	1611	0.6217	1.0	7.1170	0.6220	0.6105	0.7218
0.2823	10.0	1790	0.6449	1.0	7.1181	0.6451	0.6163	0.7114
0.1919	11.0	1969	0.6378	1.0	7.0614	0.6381	0.6162	0.7146
0.1564	12.0	2148	0.7048	1.0	7.0623	0.7051	0.6438	0.6846
0.1517	13.0	2327	0.6721	1.0	7.1234	0.6725	0.6130	0.6992

Framework versions

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