30b0526056475a6b2cb340ed84fbb08b

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

• Loss: 0.5841
• Data Size: 1.0
• Epoch Runtime: 14.3838
• Accuracy: 0.8160
• F1 Macro: 0.7917
• Rouge1: 0.8160
• Rouge2: 0.0
• Rougel: 0.8160
• Rougelsum: 0.8160

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	0.9445	0	1.9024	0.3384	0.2568	0.3379	0.0	0.3390	0.3387
No log	1	114	1.0493	0.0078	2.0931	0.3402	0.2732	0.3396	0.0	0.3402	0.3402
No log	2	228	0.8165	0.0156	2.4936	0.6651	0.3994	0.6657	0.0	0.6645	0.6651
No log	3	342	0.6571	0.0312	3.0275	0.6651	0.3994	0.6657	0.0	0.6645	0.6651
0.0233	4	456	0.6117	0.0625	3.8493	0.6963	0.5588	0.6963	0.0	0.6958	0.6958
0.0233	5	570	0.6698	0.125	4.7146	0.6875	0.4757	0.6881	0.0	0.6875	0.6875
0.0233	6	684	0.5363	0.25	6.1156	0.7966	0.7617	0.7966	0.0	0.7966	0.7972
0.1263	7	798	0.3936	0.5	8.6330	0.8384	0.8217	0.8390	0.0	0.8384	0.8390
0.3373	8.0	912	0.4000	1.0	14.7838	0.8208	0.7736	0.8208	0.0	0.8208	0.8208
0.1779	9.0	1026	0.4777	1.0	14.7321	0.8119	0.7632	0.8122	0.0	0.8119	0.8122
0.1441	10.0	1140	0.6239	1.0	14.0898	0.8249	0.7922	0.8249	0.0	0.8249	0.8255
0.1813	11.0	1254	0.5841	1.0	14.3838	0.8160	0.7917	0.8160	0.0	0.8160	0.8160

Framework versions

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