4e456a89fb272f417d95bca5d1b6e3dc

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

• Loss: 1.4712
• Data Size: 1.0
• Epoch Runtime: 63.7284
• Accuracy: 0.7564
• F1 Macro: 0.7871
• Rouge1: 0.7571
• Rouge2: 0.0
• Rougel: 0.7571
• Rougelsum: 0.7571

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	1.7270	0	4.6301	0.1854	0.1258	0.1854	0.0	0.1847	0.1854
No log	1	178	1.0282	0.0078	6.8447	0.6129	0.4390	0.6129	0.0	0.6136	0.6129
No log	2	356	0.8493	0.0156	7.9883	0.6612	0.4848	0.6619	0.0	0.6619	0.6619
No log	3	534	0.7536	0.0312	11.3403	0.7024	0.5689	0.7031	0.0	0.7031	0.7031
No log	4	712	0.6981	0.0625	15.5746	0.7351	0.5798	0.7358	0.0	0.7351	0.7351
No log	5	890	0.6445	0.125	18.3247	0.7415	0.7155	0.7422	0.0	0.7415	0.7415
0.0429	6	1068	0.6640	0.25	23.1437	0.7401	0.7377	0.7401	0.0	0.7408	0.7401
0.4673	7	1246	0.5733	0.5	38.4844	0.7848	0.8145	0.7855	0.0	0.7852	0.7841
0.3592	8.0	1424	0.6769	1.0	63.6229	0.7692	0.8024	0.7699	0.0	0.7685	0.7699
0.0946	9.0	1602	1.1242	1.0	61.3956	0.7607	0.8002	0.7614	0.0	0.7614	0.7607
0.0717	10.0	1780	1.3095	1.0	60.6049	0.7607	0.7994	0.7614	0.0	0.7614	0.7614
0.0344	11.0	1958	1.4712	1.0	63.7284	0.7564	0.7871	0.7571	0.0	0.7571	0.7571

Framework versions

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