c3334c534df77db69020adcdb8e8dee6

This model is a fine-tuned version of openai-community/gpt2 on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 1.0717
• Data Size: 1.0
• Epoch Runtime: 14.3781
• Accuracy: 0.7720
• F1 Macro: 0.7934
• Rouge1: 0.7727
• Rouge2: 0.0
• Rougel: 0.7720
• Rougelsum: 0.7727

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	8.7551	0	1.7902	0.2188	0.0749	0.2180	0.0	0.2188	0.2188
No log	1	178	3.1126	0.0078	2.8160	0.2820	0.1785	0.2827	0.0	0.2820	0.2812
No log	2	356	2.1564	0.0156	2.2757	0.3366	0.2038	0.3374	0.0	0.3374	0.3359
No log	3	534	1.3506	0.0312	2.7817	0.5398	0.4185	0.5405	0.0	0.5398	0.5398
No log	4	712	0.9207	0.0625	3.2658	0.6989	0.5513	0.6996	0.0	0.7003	0.6982
No log	5	890	0.8099	0.125	4.1053	0.7223	0.5618	0.7230	0.0	0.7237	0.7230
0.0838	6	1068	0.6785	0.25	5.3897	0.7372	0.5899	0.7379	0.0	0.7372	0.7372
0.6366	7	1246	0.6333	0.5	8.6612	0.7599	0.5997	0.7607	0.0	0.7607	0.7599
0.5248	8.0	1424	0.6111	1.0	14.9487	0.7678	0.7573	0.7685	0.0	0.7678	0.7678
0.4105	9.0	1602	0.6250	1.0	15.4161	0.7614	0.7699	0.7621	0.0	0.7621	0.7614
0.2965	10.0	1780	0.6644	1.0	17.6034	0.7592	0.7712	0.7592	0.0	0.7599	0.7592
0.2309	11.0	1958	0.7837	1.0	16.6927	0.7784	0.8046	0.7791	0.0	0.7784	0.7784
0.1439	12.0	2136	1.0717	1.0	14.3781	0.7720	0.7934	0.7727	0.0	0.7720	0.7727

Framework versions

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