scenario-kd-scr-ner-full-mdeberta-halfen_data-univner_en66

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

• Loss: 392.9234
• Precision: 0.5127
• Recall: 0.3975
• F1: 0.4478
• Accuracy: 0.9594

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: 8
• eval_batch_size: 32
• seed: 66
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
603.5902	1.28	500	532.0310	0.7568	0.0290	0.0558	0.9414
497.4866	2.55	1000	479.9597	0.2891	0.1646	0.2098	0.9463
451.2688	3.83	1500	443.6332	0.4884	0.2619	0.3410	0.9526
420.4378	5.1	2000	421.6972	0.4431	0.3789	0.4085	0.9567
400.9749	6.38	2500	408.6700	0.5306	0.3230	0.4015	0.9574
388.4506	7.65	3000	398.0555	0.5026	0.4037	0.4478	0.9596
380.4155	8.93	3500	392.9234	0.5127	0.3975	0.4478	0.9594

Framework versions

• Transformers 4.33.3
• Pytorch 2.1.1+cu121
• Datasets 2.14.5
• Tokenizers 0.13.3