99bd005b6e261d7cb3fcf9d04729cc04

This model is a fine-tuned version of studio-ousia/luke-large-lite on the nyu-mll/glue [stsb] dataset. It achieves the following results on the evaluation set:

• Loss: 0.5861
• Data Size: 1.0
• Epoch Runtime: 37.5210
• Mse: 0.5861
• Mae: 0.6055
• R2: 0.7378

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	8.2166	0	3.0424	8.2178	2.4447	-2.6761
No log	1	179	3.1348	0.0078	3.5902	3.1358	1.4890	-0.4028
No log	2	358	2.8515	0.0156	4.1603	2.8522	1.3784	-0.2759
No log	3	537	2.4379	0.0312	5.3881	2.4387	1.3355	-0.0909
No log	4	716	2.6497	0.0625	7.0827	2.6503	1.3242	-0.1856
No log	5	895	1.2511	0.125	9.9071	1.2514	0.9212	0.4402
0.1278	6	1074	0.7718	0.25	14.5046	0.7720	0.6800	0.6547
0.6089	7	1253	0.5545	0.5	21.8299	0.5545	0.5769	0.7519
0.4496	8.0	1432	0.5159	1.0	38.7442	0.5160	0.5666	0.7692
0.3086	9.0	1611	0.4559	1.0	36.9197	0.4559	0.5151	0.7961
0.246	10.0	1790	0.4288	1.0	36.5629	0.4290	0.5044	0.8081
0.2239	11.0	1969	0.4148	1.0	37.8612	0.4149	0.4949	0.8144
0.1728	12.0	2148	0.4740	1.0	37.3488	0.4741	0.5316	0.7879
0.1526	13.0	2327	0.4179	1.0	37.0386	0.4179	0.4858	0.8131
0.1474	14.0	2506	0.4627	1.0	38.4399	0.4627	0.5062	0.7930
0.1309	15.0	2685	0.4138	1.0	36.1479	0.4138	0.4862	0.8149
0.1233	16.0	2864	0.5302	1.0	37.4157	0.5301	0.5548	0.7629
0.1266	17.0	3043	0.4512	1.0	37.8095	0.4512	0.5147	0.7981
0.1012	18.0	3222	0.4328	1.0	38.1272	0.4329	0.5023	0.8064
0.0996	19.0	3401	0.5861	1.0	37.5210	0.5861	0.6055	0.7378

Framework versions

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