29b2fb76c07c099f36418ba7737b131c

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

• Loss: 0.3937
• Data Size: 1.0
• Epoch Runtime: 154.3733
• Accuracy: 0.8866
• F1 Macro: 0.8866
• Rouge1: 0.8866
• Rouge2: 0.0
• Rougel: 0.8866
• Rougelsum: 0.8868

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	0.6995	0	2.8676	0.4831	0.4239	0.4831	0.0	0.4828	0.4828
No log	1	3273	0.7037	0.0078	4.3227	0.4972	0.3385	0.4976	0.0	0.4972	0.4974
0.0109	2	6546	0.4979	0.0156	5.4515	0.7739	0.7723	0.7739	0.0	0.7737	0.7737
0.5214	3	9819	0.4226	0.0312	7.7248	0.8121	0.8115	0.8123	0.0	0.8119	0.8119
0.4554	4	13092	0.3854	0.0625	12.3875	0.8346	0.8344	0.8347	0.0	0.8346	0.8344
0.376	5	16365	0.3977	0.125	21.5760	0.8261	0.8251	0.8260	0.0	0.8259	0.8265
0.3596	6	19638	0.3566	0.25	40.2803	0.8636	0.8632	0.8638	0.0	0.8638	0.8638
0.2869	7	22911	0.3093	0.5	77.8252	0.8763	0.8761	0.8763	0.0	0.8763	0.8764
0.257	8.0	26184	0.3166	1.0	151.0262	0.8807	0.8805	0.8811	0.0	0.8809	0.8807
0.1795	9.0	29457	0.3467	1.0	158.1408	0.8849	0.8849	0.8851	0.0	0.8849	0.8849
0.1289	10.0	32730	0.4403	1.0	153.2826	0.8842	0.8841	0.8844	0.0	0.8844	0.8842
0.1383	11.0	36003	0.3937	1.0	154.3733	0.8866	0.8866	0.8866	0.0	0.8866	0.8868

Framework versions

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