3d2804b2ecbaf90df95e1b8ae09af2bf

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

• Loss: 0.9428
• Data Size: 1.0
• Epoch Runtime: 11.7767
• Accuracy: 0.7812
• F1 Macro: 0.8168
• Rouge1: 0.7812
• Rouge2: 0.0
• Rougel: 0.7812
• Rougelsum: 0.7812

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	1.6745	0	1.3944	0.2259	0.1392	0.2251	0.0	0.2259	0.2251
No log	1	178	1.3931	0.0078	1.6528	0.4276	0.2479	0.4283	0.0	0.4268	0.4276
No log	2	356	1.1945	0.0156	1.5635	0.4844	0.3175	0.4858	0.0	0.4844	0.4844
No log	3	534	1.0478	0.0312	1.7424	0.5973	0.4862	0.5980	0.0	0.5987	0.5987
No log	4	712	0.8938	0.0625	2.1720	0.6349	0.4960	0.6349	0.0	0.6357	0.6342
No log	5	890	0.8092	0.125	2.8107	0.6946	0.5266	0.6953	0.0	0.6946	0.6953
0.0603	6	1068	0.7268	0.25	4.0413	0.7138	0.5772	0.7138	0.0	0.7138	0.7138
0.6491	7	1246	0.6573	0.5	6.5657	0.7408	0.7500	0.7415	0.0	0.7415	0.7408
0.5413	8.0	1424	0.6875	1.0	11.7105	0.7337	0.7345	0.7351	0.0	0.7344	0.7337
0.4407	9.0	1602	0.6450	1.0	11.5885	0.7656	0.7938	0.7656	0.0	0.7656	0.7663
0.3626	10.0	1780	0.6046	1.0	11.3687	0.7727	0.8022	0.7734	0.0	0.7727	0.7734
0.2794	11.0	1958	0.7649	1.0	11.7194	0.7472	0.7900	0.7479	0.0	0.7472	0.7472
0.2268	12.0	2136	0.9018	1.0	11.7140	0.7670	0.8029	0.7678	0.0	0.7667	0.7678
0.1738	13.0	2314	0.9551	1.0	11.6256	0.75	0.7936	0.75	0.0	0.75	0.7507
0.1162	14.0	2492	0.9428	1.0	11.7767	0.7812	0.8168	0.7812	0.0	0.7812	0.7812

Framework versions

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