scenario-kd-pre-ner-full-mdeberta_data-univner_half66

This model is a fine-tuned version of microsoft/mdeberta-v3-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 61.3563
• Precision: 0.7742
• Recall: 0.7741
• F1: 0.7741
• Accuracy: 0.9774

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
152.339	0.5828	500	110.2190	0.5932	0.2411	0.3428	0.9360
98.8639	1.1655	1000	90.0643	0.6823	0.6661	0.6741	0.9683
84.7076	1.7483	1500	83.0368	0.7307	0.7233	0.7269	0.9733
77.9263	2.3310	2000	78.6768	0.7591	0.7112	0.7344	0.9736
73.1037	2.9138	2500	74.9915	0.7398	0.7543	0.7470	0.9758
69.158	3.4965	3000	72.5822	0.7350	0.7530	0.7439	0.9751
66.4073	4.0793	3500	70.0397	0.7805	0.7417	0.7606	0.9762
63.7231	4.6620	4000	68.5677	0.7821	0.7110	0.7449	0.9751
61.6888	5.2448	4500	66.6194	0.7556	0.7668	0.7612	0.9762
59.9681	5.8275	5000	65.3527	0.7748	0.7504	0.7624	0.9763
58.5631	6.4103	5500	64.1054	0.7718	0.7689	0.7703	0.9774
57.4916	6.9930	6000	63.4663	0.7759	0.7621	0.7689	0.9769
56.4881	7.5758	6500	62.5819	0.7680	0.7781	0.7730	0.9777
55.7575	8.1585	7000	62.0605	0.7729	0.7830	0.7779	0.9779
55.1808	8.7413	7500	61.6397	0.7711	0.7801	0.7756	0.9774
54.7861	9.3240	8000	61.4313	0.7722	0.7837	0.7779	0.9777
54.6299	9.9068	8500	61.3563	0.7742	0.7741	0.7741	0.9774

Framework versions

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