ff1a77d9459b5db6ac76e1c8163140ad

This model is a fine-tuned version of studio-ousia/luke-base-lite on the fancyzhx/dbpedia_14 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2342
• Data Size: 1.0
• Epoch Runtime: 1604.0108
• Accuracy: 0.9200
• F1 Macro: 0.9084
• Rouge1: 0.9201
• Rouge2: 0.0
• Rougel: 0.9200
• Rougelsum: 0.9201

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	2.6446	0	58.3583	0.0711	0.0095	0.0711	0.0	0.0711	0.0710
0.1242	1	17500	0.1353	0.0078	69.7075	0.9706	0.9705	0.9707	0.0	0.9706	0.9706
0.0852	2	35000	0.1019	0.0156	81.2830	0.9797	0.9799	0.9798	0.0	0.9798	0.9797
0.0657	3	52500	0.0907	0.0312	104.0434	0.9818	0.9818	0.9819	0.0	0.9818	0.9818
0.076	4	70000	0.0634	0.0625	150.5777	0.9881	0.9882	0.9881	0.0	0.9881	0.9881
0.0722	5	87500	0.0867	0.125	243.1412	0.9842	0.9842	0.9842	0.0	0.9842	0.9842
0.0713	6	105000	0.0697	0.25	428.9638	0.9866	0.9866	0.9866	0.0	0.9866	0.9866
0.0003	7	122500	0.0547	0.5	801.9799	0.9893	0.9893	0.9893	0.0	0.9893	0.9893
0.057	8.0	140000	0.0736	1.0	1533.3703	0.9875	0.9875	0.9875	0.0	0.9875	0.9875
0.0285	9.0	157500	0.0756	1.0	1561.1448	0.9884	0.9884	0.9884	0.0	0.9884	0.9884
0.8809	10.0	175000	0.8604	1.0	1536.7656	0.5994	0.5252	0.5995	0.0	0.5995	0.5996
0.1438	11.0	192500	0.2342	1.0	1604.0108	0.9200	0.9084	0.9201	0.0	0.9200	0.9201

Framework versions

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