WITHINAPPS_NDD-pagekit_test-content-LongFormer

This model is a fine-tuned version of allenai/longformer-base-4096 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1262
• Accuracy: 0.8643
• F1: 0.8651
• Precision: 0.8941
• Recall: 0.8643

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: 2e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.1837	1.0	973	0.1281	0.8469	0.8474	0.8841	0.8469
0.1508	2.0	1946	0.1247	0.8561	0.8571	0.8792	0.8561
0.1374	3.0	2919	0.1459	0.8571	0.8581	0.8773	0.8571
0.1353	4.0	3892	0.1400	0.8607	0.8616	0.8860	0.8607
0.1313	5.0	4865	0.1416	0.8587	0.8596	0.8831	0.8587
0.1281	6.0	5838	0.1195	0.8643	0.8652	0.8907	0.8643
0.1246	7.0	6811	0.1373	0.8623	0.8632	0.8866	0.8623
0.1261	8.0	7784	0.1193	0.8618	0.8626	0.8895	0.8618
0.1168	9.0	8757	0.1223	0.8643	0.8650	0.8955	0.8643
0.1184	10.0	9730	0.1232	0.8638	0.8648	0.8817	0.8638
0.1195	11.0	10703	0.1192	0.8664	0.8671	0.8971	0.8664
0.1176	12.0	11676	0.1294	0.8654	0.8661	0.8956	0.8654
0.1126	13.0	12649	0.1199	0.8602	0.8612	0.8766	0.8602
0.115	14.0	13622	0.1200	0.8633	0.8641	0.8931	0.8633
0.1102	15.0	14595	0.1213	0.8654	0.8661	0.8961	0.8654
0.1089	16.0	15568	0.1229	0.8628	0.8636	0.8902	0.8628
0.112	17.0	16541	0.1199	0.8669	0.8676	0.8983	0.8669
0.1096	18.0	17514	0.1242	0.8669	0.8676	0.8983	0.8669
0.1091	19.0	18487	0.1262	0.8643	0.8651	0.8941	0.8643
0.108	20.0	19460	0.1262	0.8643	0.8651	0.8941	0.8643

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1