ce744b05851cf4e216fbb8e2b1207eee

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

• Loss: 0.7308
• Data Size: 1.0
• Epoch Runtime: 319.3708
• Accuracy: 0.7793
• F1 Macro: 0.7792
• Rouge1: 0.7793
• Rouge2: 0.0
• Rougel: 0.7791
• Rougelsum: 0.7795

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	1.1005	0	3.0779	0.3542	0.1753	0.3541	0.0	0.3542	0.3540
1.0591	1	12271	0.9117	0.0078	5.7548	0.5910	0.5890	0.5912	0.0	0.5912	0.5914
0.8843	2	24542	0.8494	0.0156	8.2110	0.6286	0.6208	0.6284	0.0	0.6290	0.6289
0.8003	3	36813	0.7587	0.0312	13.4061	0.6711	0.6670	0.6710	0.0	0.6711	0.6710
0.7269	4	49084	0.6992	0.0625	23.3519	0.7078	0.7085	0.7076	0.0	0.7080	0.7078
0.6138	5	61355	0.6148	0.125	42.4668	0.7472	0.7454	0.7474	0.0	0.7475	0.7473
0.6038	6	73626	0.6194	0.25	81.9287	0.7501	0.7505	0.7499	0.0	0.7502	0.75
0.5292	7	85897	0.5717	0.5	160.9253	0.7718	0.7707	0.7714	0.0	0.7720	0.7719
0.4832	8.0	98168	0.5510	1.0	322.0108	0.7831	0.7834	0.7832	0.0	0.7830	0.7831
0.3897	9.0	110439	0.5706	1.0	316.3502	0.7876	0.7869	0.7874	0.0	0.7876	0.7877
0.3573	10.0	122710	0.5990	1.0	318.3922	0.7851	0.7828	0.7848	0.0	0.7851	0.7849
0.2757	11.0	134981	0.6851	1.0	317.6002	0.7810	0.7814	0.7809	0.0	0.7811	0.7809
0.2328	12.0	147252	0.7308	1.0	319.3708	0.7793	0.7792	0.7793	0.0	0.7791	0.7795

Framework versions

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