764e9c16bd5e36e8d38a1d3a71d4f337

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

• Loss: 0.8697
• Data Size: 1.0
• Epoch Runtime: 6.9310
• Accuracy: 0.7614
• F1 Macro: 0.7976
• Rouge1: 0.7614
• Rouge2: 0.0
• Rougel: 0.7617
• Rougelsum: 0.7614

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	1.6877	0	1.1516	0.2408	0.0930	0.2408	0.0	0.2408	0.2408
No log	1	178	1.4572	0.0078	1.3273	0.4119	0.2730	0.4126	0.0	0.4126	0.4112
No log	2	356	1.3424	0.0156	1.2503	0.4624	0.2968	0.4631	0.0	0.4631	0.4616
No log	3	534	1.0930	0.0312	1.2713	0.6328	0.5020	0.6335	0.0	0.6328	0.6328
No log	4	712	0.8348	0.0625	1.5452	0.6875	0.5035	0.6882	0.0	0.6875	0.6875
No log	5	890	0.7604	0.125	1.8696	0.7081	0.5430	0.7088	0.0	0.7088	0.7088
0.0579	6	1068	0.6835	0.25	2.6009	0.7195	0.5746	0.7202	0.0	0.7195	0.7195
0.6188	7	1246	0.6388	0.5	4.0470	0.7479	0.7277	0.7493	0.0	0.7479	0.7479
0.5303	8.0	1424	0.5805	1.0	7.1521	0.7741	0.7829	0.7741	0.0	0.7749	0.7741
0.4377	9.0	1602	0.5852	1.0	6.9961	0.7685	0.8004	0.7692	0.0	0.7692	0.7692
0.3168	10.0	1780	0.7057	1.0	6.9122	0.7564	0.7953	0.7571	0.0	0.7564	0.7571
0.2304	11.0	1958	0.7780	1.0	6.9348	0.7685	0.8075	0.7685	0.0	0.7685	0.7685
0.1476	12.0	2136	0.8697	1.0	6.9310	0.7614	0.7976	0.7614	0.0	0.7617	0.7614

Framework versions

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