029ea5a9973ca7672d4f63af802a9614

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

• Loss: 0.4841
• Data Size: 1.0
• Epoch Runtime: 36.7842
• Mse: 0.4842
• Mae: 0.5275
• R2: 0.7834

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	8.0694	0	2.8893	8.0706	2.4163	-2.6103
No log	1	179	4.0866	0.0078	3.7698	4.0877	1.6523	-0.8286
No log	2	358	3.7891	0.0156	3.9160	3.7901	1.6106	-0.6955
No log	3	537	2.8680	0.0312	5.5733	2.8685	1.4032	-0.2832
No log	4	716	2.4632	0.0625	7.6556	2.4641	1.3432	-0.1023
No log	5	895	2.2990	0.125	10.0617	2.2998	1.2733	-0.0288
0.1899	6	1074	0.8564	0.25	14.8123	0.8569	0.7272	0.6167
0.7716	7	1253	0.6077	0.5	24.3844	0.6081	0.6279	0.7280
0.5351	8.0	1432	0.4998	1.0	40.3381	0.5001	0.5700	0.7763
0.3326	9.0	1611	0.4956	1.0	36.5045	0.4958	0.5464	0.7782
0.2897	10.0	1790	0.5737	1.0	37.7494	0.5737	0.5835	0.7434
0.224	11.0	1969	0.6626	1.0	37.5533	0.6625	0.6259	0.7036
0.1742	12.0	2148	0.6232	1.0	36.4729	0.6233	0.5975	0.7212
0.164	13.0	2327	0.4797	1.0	37.0413	0.4797	0.5247	0.7854
0.1235	14.0	2506	0.5450	1.0	36.2418	0.5450	0.5508	0.7562
0.1068	15.0	2685	0.5316	1.0	36.4645	0.5316	0.5460	0.7622
0.102	16.0	2864	0.4720	1.0	36.6696	0.4720	0.5192	0.7889
0.0962	17.0	3043	0.4488	1.0	36.5595	0.4489	0.5036	0.7992
0.0788	18.0	3222	0.4922	1.0	37.7005	0.4922	0.5357	0.7798
0.0755	19.0	3401	0.4559	1.0	36.4642	0.4560	0.5075	0.7960
0.0791	20.0	3580	0.4919	1.0	36.6052	0.4920	0.5331	0.7799
0.0752	21.0	3759	0.4841	1.0	36.7842	0.4842	0.5275	0.7834

Framework versions

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