d5a2c91be4d64cbafd816d8b87ec80bf

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

• Loss: 0.3242
• Data Size: 1.0
• Epoch Runtime: 231.7537
• Accuracy: 0.8895
• F1 Macro: 0.8895
• Rouge1: 0.8899
• Rouge2: 0.0
• Rougel: 0.8895
• Rougelsum: 0.8893

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	0.7021	0	4.4355	0.4943	0.3308	0.4947	0.0	0.4943	0.4944
No log	1	3273	0.6567	0.0078	6.4856	0.7829	0.7828	0.7829	0.0	0.7825	0.7825
0.0102	2	6546	0.5043	0.0156	8.4778	0.7739	0.7655	0.7740	0.0	0.7737	0.7735
0.4611	3	9819	0.4269	0.0312	12.0791	0.8281	0.8275	0.8281	0.0	0.8283	0.8279
0.4312	4	13092	0.3636	0.0625	19.1919	0.8546	0.8545	0.8547	0.0	0.8546	0.8540
0.3594	5	16365	0.3498	0.125	33.2699	0.8711	0.8711	0.8713	0.0	0.8710	0.8708
0.3787	6	19638	0.3046	0.25	61.7871	0.8814	0.8814	0.8814	0.0	0.8815	0.8812
0.3074	7	22911	0.2846	0.5	118.4497	0.8857	0.8855	0.8858	0.0	0.8858	0.8858
0.2926	8.0	26184	0.3357	1.0	232.1482	0.8895	0.8894	0.8897	0.0	0.8895	0.8895
0.2215	9.0	29457	0.3058	1.0	231.8773	0.8932	0.8932	0.8932	0.0	0.8932	0.8930
0.2326	10.0	32730	0.3017	1.0	230.3343	0.8881	0.8879	0.8881	0.0	0.8882	0.8877
0.2008	11.0	36003	0.3242	1.0	231.7537	0.8895	0.8895	0.8899	0.0	0.8895	0.8893

Framework versions

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