d88f8a24204e041736ed668fc9f3d58d

This model is a fine-tuned version of distilbert/distilgpt2 on the nyu-mll/glue [mrpc] dataset. It achieves the following results on the evaluation set:

• Loss: 0.9839
• Data Size: 1.0
• Epoch Runtime: 6.7462
• Accuracy: 0.7624
• F1 Macro: 0.7029
• Rouge1: 0.7624
• Rouge2: 0.0
• Rougel: 0.7624
• Rougelsum: 0.7630

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	3.0948	0	1.5038	0.3496	0.2830	0.3491	0.0	0.3496	0.3496
No log	1	114	1.2284	0.0078	2.8503	0.4817	0.4746	0.4811	0.0	0.4817	0.4817
No log	2	228	1.1148	0.0156	2.0155	0.5702	0.4326	0.5702	0.0	0.5708	0.5696
No log	3	342	0.7706	0.0312	2.2081	0.5991	0.4360	0.5996	0.0	0.5991	0.5991
0.0346	4	456	0.7865	0.0625	2.3430	0.4522	0.4493	0.4517	0.0	0.4519	0.4517
0.0346	5	570	0.6624	0.125	2.4761	0.6792	0.4774	0.6792	0.0	0.6792	0.6792
0.0346	6	684	0.5813	0.25	2.9420	0.7040	0.5695	0.7040	0.0	0.7034	0.7040
0.1516	7	798	0.5438	0.5	4.0452	0.7335	0.6596	0.7335	0.0	0.7335	0.7335
0.5075	8.0	912	0.5269	1.0	6.4291	0.7453	0.6571	0.7459	0.0	0.7459	0.7459
0.4486	9.0	1026	0.5587	1.0	6.2537	0.7689	0.7052	0.7689	0.0	0.7689	0.7689
0.3363	10.0	1140	0.5860	1.0	6.3052	0.7642	0.7348	0.7636	0.0	0.7647	0.7647
0.2245	11.0	1254	0.7463	1.0	6.3571	0.7588	0.6922	0.7588	0.0	0.7588	0.7594
0.158	12.0	1368	0.9839	1.0	6.7462	0.7624	0.7029	0.7624	0.0	0.7624	0.7630

Framework versions

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