scenario-kd-po-ner-half-mdeberta_data-univner_full66

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

• Loss: 72.5559
• Precision: 0.6792
• Recall: 0.6229
• F1: 0.6498
• Accuracy: 0.9661

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
171.7606	0.29	500	133.7154	0.0	0.0	0.0	0.9241
125.3819	0.58	1000	118.8982	0.0	0.0	0.0	0.9241
113.475	0.87	1500	108.9966	0.1272	0.0092	0.0172	0.9252
104.3223	1.16	2000	100.9579	0.2417	0.1160	0.1568	0.9331
96.8596	1.46	2500	95.4198	0.2974	0.1661	0.2131	0.9368
92.6352	1.75	3000	90.9541	0.4272	0.2581	0.3218	0.9426
88.4087	2.04	3500	87.7702	0.5057	0.3189	0.3911	0.9466
84.9655	2.33	4000	85.1105	0.5218	0.3844	0.4427	0.9507
82.6755	2.62	4500	83.3483	0.5638	0.3848	0.4574	0.9511
80.9047	2.91	5000	81.5072	0.5652	0.4416	0.4958	0.9547
78.7895	3.2	5500	80.3758	0.5558	0.5207	0.5377	0.9574
77.711	3.49	6000	79.3018	0.5889	0.5400	0.5634	0.9591
76.5548	3.78	6500	78.2521	0.6095	0.5271	0.5653	0.9596
75.9175	4.08	7000	77.5573	0.6196	0.5468	0.5809	0.9602
74.7471	4.37	7500	77.0193	0.6033	0.5641	0.5831	0.9603
73.9283	4.66	8000	76.3540	0.6221	0.5842	0.6025	0.9622
73.7326	4.95	8500	75.7315	0.6342	0.5937	0.6133	0.9631
73.0742	5.24	9000	75.2493	0.6512	0.5800	0.6136	0.9634
72.4297	5.53	9500	75.0674	0.6474	0.5933	0.6191	0.9633
71.8295	5.82	10000	74.5560	0.6396	0.6154	0.6273	0.9642
71.572	6.11	10500	74.2151	0.6623	0.6081	0.6341	0.9645
71.1983	6.4	11000	73.9930	0.6776	0.5926	0.6322	0.9649
71.0043	6.69	11500	73.8112	0.6733	0.5979	0.6333	0.9644
70.7001	6.99	12000	73.5677	0.6763	0.6066	0.6395	0.9653
70.4901	7.28	12500	73.4664	0.6802	0.5918	0.6329	0.9647
70.1377	7.57	13000	73.1932	0.6639	0.6272	0.6450	0.9661
69.8947	7.86	13500	73.0598	0.6804	0.6211	0.6494	0.9663
69.7927	8.15	14000	72.9965	0.6746	0.6242	0.6484	0.9658
69.5853	8.44	14500	72.8758	0.6715	0.6188	0.6441	0.9655
69.5151	8.73	15000	72.6126	0.6782	0.6275	0.6518	0.9664
69.2025	9.02	15500	72.7029	0.6640	0.6220	0.6423	0.9654
69.2257	9.31	16000	72.5754	0.6779	0.6236	0.6496	0.9660
69.145	9.61	16500	72.4798	0.6744	0.6203	0.6462	0.9665
69.0684	9.9	17000	72.5559	0.6792	0.6229	0.6498	0.9661

Framework versions

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