5edafecd827b8f8cb82b3082c3572bdb

This model is a fine-tuned version of albert/albert-xxlarge-v1 on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 1.2629
• Data Size: 1.0
• Epoch Runtime: 26.8143
• Accuracy: 0.7472
• F1 Macro: 0.7952
• Rouge1: 0.7464
• Rouge2: 0.0
• Rougel: 0.7479
• Rougelsum: 0.7472

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	1.8987	0	2.2973	0.2678	0.1437	0.2670	0.0	0.2670	0.2678
No log	1	178	1.6973	0.0078	3.3852	0.2237	0.1187	0.2244	0.0	0.2237	0.2230
No log	2	356	1.5283	0.0156	2.8877	0.3068	0.1411	0.3068	0.0	0.3061	0.3061
No log	3	534	1.3080	0.0312	3.4598	0.4560	0.3380	0.4567	0.0	0.4567	0.4560
No log	4	712	0.9668	0.0625	4.3074	0.6477	0.4462	0.6484	0.0	0.6484	0.6470
No log	5	890	0.8541	0.125	5.6995	0.6889	0.5250	0.6896	0.0	0.6896	0.6889
0.0653	6	1068	0.6460	0.25	8.7033	0.7464	0.5951	0.7479	0.0	0.7472	0.7472
0.5924	7	1246	0.5770	0.5	14.6222	0.7749	0.8043	0.7749	0.0	0.7749	0.7756
0.4992	8.0	1424	0.6034	1.0	27.0950	0.7699	0.7905	0.7706	0.0	0.7706	0.7699
0.359	9.0	1602	0.6740	1.0	26.8618	0.7763	0.8050	0.7770	0.0	0.7763	0.7763
0.1543	10.0	1780	0.7909	1.0	26.8134	0.7756	0.8062	0.7763	0.0	0.7763	0.7756
0.1077	11.0	1958	1.2629	1.0	26.8143	0.7472	0.7952	0.7464	0.0	0.7479	0.7472

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.1