1bcc5bb289b49178c8102d291edd5634

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

• Loss: 0.4738
• Data Size: 1.0
• Epoch Runtime: 15.3219
• Mse: 0.4740
• Mae: 0.5264
• R2: 0.7879

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	6.9295	0	1.6349	6.9307	2.2025	-2.1004
No log	1	179	5.4552	0.0078	2.0674	5.4564	1.9334	-1.4408
No log	2	358	2.1803	0.0156	2.0406	2.1812	1.2750	0.0243
No log	3	537	2.1721	0.0312	2.4043	2.1730	1.2713	0.0279
No log	4	716	2.3324	0.0625	2.9166	2.3332	1.2777	-0.0437
No log	5	895	1.5504	0.125	3.6610	1.5505	1.0381	0.3064
0.1407	6	1074	0.7633	0.25	5.2091	0.7634	0.6882	0.6585
0.6503	7	1253	0.5737	0.5	8.4262	0.5740	0.6077	0.7432
0.5047	8.0	1432	0.4593	1.0	15.0540	0.4596	0.5177	0.7944
0.3605	9.0	1611	0.4260	1.0	14.5109	0.4261	0.4844	0.8094
0.2824	10.0	1790	0.4959	1.0	14.9747	0.4961	0.5516	0.7781
0.2249	11.0	1969	0.5420	1.0	14.7415	0.5420	0.5667	0.7575
0.1729	12.0	2148	0.4392	1.0	15.0047	0.4394	0.4925	0.8035
0.1521	13.0	2327	0.4738	1.0	15.3219	0.4740	0.5264	0.7879

Framework versions

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