9235b6d74f72fe6a0ca31fbb2c07b9ea

This model is a fine-tuned version of google-bert/bert-large-uncased-whole-word-masking-finetuned-squad on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 1.0952
• Data Size: 1.0
• Epoch Runtime: 20.1387
• Accuracy: 0.7812
• F1 Macro: 0.8144
• Rouge1: 0.7820
• Rouge2: 0.0
• Rougel: 0.7820
• 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	1.8097	0	1.6700	0.0185	0.0073	0.0185	0.0	0.0185	0.0185
No log	1	178	1.4957	0.0078	2.2245	0.2401	0.0775	0.2401	0.0	0.2408	0.2393
No log	2	356	1.3334	0.0156	2.2607	0.4595	0.3261	0.4602	0.0	0.4588	0.4595
No log	3	534	1.0233	0.0312	3.0793	0.6087	0.4775	0.6087	0.0	0.6094	0.6080
No log	4	712	0.8250	0.0625	4.1178	0.7095	0.5598	0.7109	0.0	0.7095	0.7095
No log	5	890	0.8420	0.125	6.0440	0.6932	0.4848	0.6939	0.0	0.6939	0.6932
0.0565	6	1068	0.6822	0.25	8.1848	0.7436	0.6871	0.7450	0.0	0.7436	0.7429
0.6061	7	1246	0.6300	0.5	11.7259	0.7670	0.7912	0.7678	0.0	0.7678	0.7670
0.47	8.0	1424	0.6061	1.0	20.2008	0.7699	0.7623	0.7706	0.0	0.7706	0.7699
0.3081	9.0	1602	0.7404	1.0	20.4169	0.7777	0.7908	0.7784	0.0	0.7777	0.7777
0.2051	10.0	1780	0.9964	1.0	20.1384	0.7493	0.7951	0.75	0.0	0.75	0.7493
0.1443	11.0	1958	1.1498	1.0	19.9721	0.7571	0.8000	0.7571	0.0	0.7571	0.7571
0.1242	12.0	2136	1.0952	1.0	20.1387	0.7812	0.8144	0.7820	0.0	0.7820	0.7812

Framework versions

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