af50a2bdac15102ab012de8944d739c8

This model is a fine-tuned version of google-bert/bert-base-cased on the fancyzhx/dbpedia_14 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2948
• Data Size: 1.0
• Epoch Runtime: 830.6057
• Accuracy: 0.9141
• F1 Macro: 0.8926
• Rouge1: 0.9141
• Rouge2: 0.0
• Rougel: 0.9142
• Rougelsum: 0.9141

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	2.7061	0	29.3340	0.0714	0.0101	0.0715	0.0	0.0715	0.0714
0.1342	1	17500	0.0872	0.0078	35.5730	0.9821	0.9821	0.9822	0.0	0.9821	0.9821
0.0536	2	35000	0.0790	0.0156	41.5098	0.9856	0.9857	0.9857	0.0	0.9856	0.9857
0.0422	3	52500	0.0707	0.0312	54.2258	0.9857	0.9857	0.9857	0.0	0.9856	0.9857
0.0793	4	70000	0.0621	0.0625	78.4774	0.9872	0.9872	0.9872	0.0	0.9871	0.9871
0.0494	5	87500	0.0711	0.125	128.4128	0.9858	0.9858	0.9858	0.0	0.9858	0.9858
0.0614	6	105000	0.0528	0.25	227.0372	0.9892	0.9892	0.9892	0.0	0.9892	0.9892
0.0004	7	122500	0.0505	0.5	425.1327	0.9878	0.9877	0.9878	0.0	0.9877	0.9878
0.0531	8.0	140000	0.0684	1.0	819.0964	0.9879	0.9879	0.9879	0.0	0.9879	0.9880
0.0493	9.0	157500	0.0697	1.0	833.8277	0.9876	0.9877	0.9876	0.0	0.9876	0.9876
0.0932	10.0	175000	0.4105	1.0	840.5211	0.9097	0.8892	0.9097	0.0	0.9097	0.9098
0.1038	11.0	192500	0.2948	1.0	830.6057	0.9141	0.8926	0.9141	0.0	0.9142	0.9141

Framework versions

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