scenario-kd-po-ner-full-xlmr_data-univner_half44

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: 53.5855
• Precision: 0.7926
• Recall: 0.7941
• F1: 0.7934
• 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: 8
• eval_batch_size: 32
• seed: 44
• 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
94.1316	0.5828	500	77.7639	0.7573	0.7254	0.7410	0.9749
69.1816	1.1655	1000	70.0861	0.7700	0.7553	0.7626	0.9771
62.5325	1.7483	1500	66.1010	0.7697	0.7728	0.7712	0.9774
58.8032	2.3310	2000	63.2843	0.7722	0.7813	0.7767	0.9781
55.8439	2.9138	2500	61.3899	0.7711	0.7839	0.7774	0.9777
53.6386	3.4965	3000	59.5744	0.7829	0.7804	0.7816	0.9782
51.8854	4.0793	3500	58.4745	0.7896	0.7831	0.7864	0.9784
50.3704	4.6620	4000	57.3648	0.7888	0.7917	0.7902	0.9787
49.2945	5.2448	4500	56.3673	0.8003	0.7807	0.7904	0.9787
48.3678	5.8275	5000	55.7695	0.7906	0.7840	0.7873	0.9787
47.4721	6.4103	5500	55.1454	0.7836	0.7964	0.7900	0.9792
46.9783	6.9930	6000	54.6410	0.7931	0.7976	0.7953	0.9790
46.3896	7.5758	6500	54.2132	0.8004	0.7902	0.7953	0.9792
45.895	8.1585	7000	53.9535	0.7906	0.7945	0.7925	0.9792
45.6796	8.7413	7500	53.7738	0.7918	0.7895	0.7906	0.9788
45.4159	9.3240	8000	53.6266	0.7904	0.7950	0.7927	0.9793
45.3274	9.9068	8500	53.5855	0.7926	0.7941	0.7934	0.9792

Framework versions

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