scenario-non-kd-po-ner-full-mdeberta_data-univner_en55

This model is a fine-tuned version of haryoaw/scenario-TCR-NER_data-univner_en on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1837
• Precision: 0.7462
• Recall: 0.7547
• F1: 0.7504
• Accuracy: 0.9792

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 55
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.1134	1.2755	500	0.1161	0.5285	0.6718	0.5916	0.9631
0.0431	2.5510	1000	0.0940	0.6331	0.7609	0.6911	0.9744
0.0222	3.8265	1500	0.0978	0.6914	0.7536	0.7211	0.9774
0.0127	5.1020	2000	0.1162	0.6961	0.7516	0.7227	0.9771
0.0079	6.3776	2500	0.1259	0.6670	0.7650	0.7126	0.9756
0.0059	7.6531	3000	0.1308	0.6906	0.7464	0.7174	0.9767
0.0039	8.9286	3500	0.1348	0.7239	0.7464	0.7350	0.9788
0.0035	10.2041	4000	0.1420	0.6892	0.7692	0.7270	0.9775
0.0021	11.4796	4500	0.1484	0.7104	0.7516	0.7304	0.9778
0.002	12.7551	5000	0.1477	0.7158	0.7795	0.7463	0.9790
0.0016	14.0306	5500	0.1590	0.7232	0.7464	0.7346	0.9782
0.001	15.3061	6000	0.1585	0.7182	0.7598	0.7384	0.9785
0.0012	16.5816	6500	0.1622	0.7536	0.7505	0.7521	0.9796
0.0008	17.8571	7000	0.1703	0.7521	0.7505	0.7513	0.9800
0.0008	19.1327	7500	0.1686	0.7226	0.7578	0.7398	0.9782
0.0005	20.4082	8000	0.1803	0.7099	0.7598	0.7340	0.9786
0.0006	21.6837	8500	0.1718	0.7307	0.7640	0.7470	0.9786
0.0007	22.9592	9000	0.1826	0.7369	0.7422	0.7396	0.9790
0.0005	24.2347	9500	0.1811	0.7339	0.7650	0.7491	0.9787
0.0004	25.5102	10000	0.1820	0.7111	0.7619	0.7356	0.9777
0.0003	26.7857	10500	0.1840	0.7218	0.7681	0.7442	0.9781
0.0003	28.0612	11000	0.1854	0.7477	0.7516	0.7496	0.9794
0.0002	29.3367	11500	0.1837	0.7462	0.7547	0.7504	0.9792

Framework versions

• Transformers 4.44.2
• Pytorch 2.1.1+cu121
• Datasets 2.14.5
• Tokenizers 0.19.1