f622b727b75817b8f1423db832627c75

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

• Loss: 0.4831
• Data Size: 1.0
• Epoch Runtime: 874.4305
• Accuracy: 0.8283
• F1 Macro: 0.8277
• Rouge1: 0.8280
• Rouge2: 0.0
• Rougel: 0.8284
• Rougelsum: 0.8284

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	Accuracy	F1 Macro	Rouge1	Rouge2	Rougel	Rougelsum
No log	0	0	1.1002	0	7.5537	0.3545	0.1745	0.3544	0.0	0.3545	0.3543
1.0914	1	12271	0.7685	0.0078	14.5055	0.6877	0.6848	0.6874	0.0	0.6880	0.6878
0.7063	2	24542	0.6647	0.0156	20.8705	0.7465	0.7442	0.7467	0.0	0.7465	0.7463
0.5861	3	36813	0.5379	0.0312	34.4086	0.7895	0.7881	0.7895	0.0	0.7896	0.7896
0.5811	4	49084	0.4942	0.0625	61.7365	0.8128	0.8125	0.8126	0.0	0.8130	0.8128
0.4964	5	61355	0.4725	0.125	114.7542	0.8177	0.8159	0.8175	0.0	0.8179	0.8178
0.5248	6	73626	0.4828	0.25	217.8939	0.8136	0.8128	0.8133	0.0	0.8137	0.8135
0.4341	7	85897	0.4580	0.5	429.1974	0.8278	0.8262	0.8279	0.0	0.8280	0.8279
0.4275	8.0	98168	0.4558	1.0	871.1783	0.8349	0.8338	0.8346	0.0	0.8350	0.8350
0.3706	9.0	110439	0.5000	1.0	880.3824	0.8281	0.8274	0.8280	0.0	0.8279	0.8280
0.3209	10.0	122710	0.4714	1.0	873.8713	0.8278	0.8267	0.8276	0.0	0.8278	0.8280
0.3122	11.0	134981	0.5268	1.0	875.4766	0.8255	0.8255	0.8254	0.0	0.8256	0.8254
0.3311	12.0	147252	0.4831	1.0	874.4305	0.8283	0.8277	0.8280	0.0	0.8284	0.8284

Framework versions

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