91dd4848df1dcc6516aebde9781fe96a

This model is a fine-tuned version of albert/albert-base-v2 on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 0.8219
• Data Size: 1.0
• Epoch Runtime: 7.7663
• Accuracy: 0.7656
• F1 Macro: 0.8024
• Rouge1: 0.7656
• Rouge2: 0.0
• Rougel: 0.7656
• Rougelsum: 0.7660

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	1.5855	0	1.1480	0.2528	0.1324	0.2536	0.0	0.2528	0.2521
No log	1	178	1.5183	0.0078	1.3067	0.2464	0.0872	0.2464	0.0	0.2472	0.2464
No log	2	356	1.2755	0.0156	1.3029	0.5114	0.3377	0.5121	0.0	0.5121	0.5107
No log	3	534	1.0574	0.0312	1.3808	0.5817	0.4498	0.5817	0.0	0.5824	0.5817
No log	4	712	0.8496	0.0625	1.6570	0.6683	0.4776	0.6690	0.0	0.6690	0.6683
No log	5	890	0.8338	0.125	2.0610	0.6804	0.5015	0.6804	0.0	0.6811	0.6804
0.06	6	1068	0.7227	0.25	2.8779	0.7202	0.5639	0.7209	0.0	0.7212	0.7209
0.6694	7	1246	0.6656	0.5	4.5552	0.7543	0.7774	0.7550	0.0	0.7543	0.7543
0.5675	8.0	1424	0.5886	1.0	7.8715	0.7614	0.7869	0.7614	0.0	0.7621	0.7614
0.4731	9.0	1602	0.5916	1.0	7.8089	0.7678	0.8056	0.7678	0.0	0.7685	0.7685
0.3789	10.0	1780	0.5807	1.0	7.8636	0.7812	0.8167	0.7820	0.0	0.7812	0.7820
0.3058	11.0	1958	0.7981	1.0	7.7563	0.7216	0.7655	0.7216	0.0	0.7216	0.7216
0.2623	12.0	2136	0.8125	1.0	7.8404	0.7578	0.7909	0.7585	0.0	0.7585	0.7571
0.1936	13.0	2314	0.8475	1.0	7.7562	0.7557	0.8008	0.7557	0.0	0.7557	0.7564
0.1767	14.0	2492	0.8219	1.0	7.7663	0.7656	0.8024	0.7656	0.0	0.7656	0.7660

Framework versions

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