66c833920a2cacbd5f4c1f2e4cb3501a

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

• Loss: 0.6483
• Data Size: 1.0
• Epoch Runtime: 583.0881
• Accuracy: 0.7954
• F1 Macro: 0.7955
• Rouge1: 0.7952
• Rouge2: 0.0
• Rougel: 0.7956
• Rougelsum: 0.7955

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	1.1380	0	4.7724	0.3150	0.2096	0.3148	0.0	0.3148	0.3151
1.0722	1	12271	0.9158	0.0078	9.1687	0.5852	0.5847	0.5852	0.0	0.5852	0.5855
0.8206	2	24542	0.7758	0.0156	14.3380	0.6679	0.6594	0.6681	0.0	0.6677	0.6680
0.6995	3	36813	0.6654	0.0312	21.8765	0.7227	0.7191	0.7225	0.0	0.7225	0.7228
0.6549	4	49084	0.6199	0.0625	40.0906	0.7469	0.7471	0.7468	0.0	0.7470	0.7469
0.56	5	61355	0.5797	0.125	77.1214	0.7619	0.7607	0.7618	0.0	0.7619	0.7621
0.5762	6	73626	0.5656	0.25	147.7526	0.7787	0.7785	0.7785	0.0	0.7787	0.7785
0.4822	7	85897	0.5480	0.5	285.2704	0.7862	0.7852	0.7862	0.0	0.7862	0.7863
0.4611	8.0	98168	0.5689	1.0	574.3775	0.7896	0.7904	0.7896	0.0	0.7893	0.7895
0.3584	9.0	110439	0.6012	1.0	582.0234	0.7926	0.7919	0.7924	0.0	0.7925	0.7923
0.3155	10.0	122710	0.5876	1.0	577.8500	0.7927	0.7910	0.7927	0.0	0.7927	0.7928
0.2492	11.0	134981	0.6483	1.0	583.0881	0.7954	0.7955	0.7952	0.0	0.7956	0.7955

Framework versions

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