README.md

metadata

base_model: microsoft/mdeberta-v3-base
library_name: transformers
license: mit
metrics:
  - precision
  - recall
  - f1
  - accuracy
tags:
  - generated_from_trainer
model-index:
  - name: scenario-non-kd-scr-ner-half-mdeberta_data-univner_full55
    results: []

scenario-non-kd-scr-ner-half-mdeberta_data-univner_full55

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: 0.2603
• Precision: 0.6078
• Recall: 0.5840
• F1: 0.5957
• Accuracy: 0.9607

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
0.3519	0.2910	500	0.2816	0.3080	0.1195	0.1722	0.9290
0.2346	0.5821	1000	0.2141	0.3649	0.2431	0.2918	0.9381
0.1749	0.8731	1500	0.1763	0.4217	0.3763	0.3977	0.9469
0.1347	1.1641	2000	0.1684	0.4579	0.4136	0.4347	0.9507
0.1095	1.4552	2500	0.1560	0.4890	0.5004	0.4946	0.9535
0.1004	1.7462	3000	0.1529	0.5150	0.5223	0.5186	0.9557
0.0893	2.0373	3500	0.1536	0.5390	0.5421	0.5405	0.9573
0.0687	2.3283	4000	0.1621	0.5659	0.5190	0.5414	0.9578
0.0651	2.6193	4500	0.1575	0.5644	0.5419	0.5529	0.9585
0.0633	2.9104	5000	0.1500	0.5622	0.5718	0.5670	0.9598
0.0495	3.2014	5500	0.1585	0.5754	0.5763	0.5758	0.9600
0.0441	3.4924	6000	0.1694	0.5877	0.5646	0.5759	0.9600
0.0439	3.7835	6500	0.1642	0.5763	0.5881	0.5821	0.9602
0.04	4.0745	7000	0.1719	0.5795	0.5920	0.5857	0.9604
0.0298	4.3655	7500	0.1847	0.5904	0.5561	0.5727	0.9604
0.0328	4.6566	8000	0.1730	0.6106	0.5944	0.6024	0.9614
0.0312	4.9476	8500	0.1730	0.5697	0.6041	0.5864	0.9595
0.0223	5.2386	9000	0.1855	0.5739	0.6086	0.5907	0.9603
0.0211	5.5297	9500	0.1962	0.6010	0.5703	0.5853	0.9607
0.0225	5.8207	10000	0.1951	0.6058	0.5778	0.5915	0.9609
0.0208	6.1118	10500	0.2037	0.5960	0.5905	0.5932	0.9604
0.0159	6.4028	11000	0.2101	0.5885	0.5812	0.5848	0.9600
0.0167	6.6938	11500	0.2092	0.5946	0.5852	0.5899	0.9605
0.0167	6.9849	12000	0.2103	0.5901	0.5826	0.5863	0.9602
0.0113	7.2759	12500	0.2245	0.5868	0.5888	0.5878	0.9601
0.0114	7.5669	13000	0.2301	0.6030	0.5799	0.5912	0.9610
0.0122	7.8580	13500	0.2336	0.6091	0.5641	0.5858	0.9605
0.0115	8.1490	14000	0.2328	0.5922	0.5868	0.5895	0.9604
0.0087	8.4400	14500	0.2393	0.6049	0.5719	0.5880	0.9607
0.0094	8.7311	15000	0.2455	0.5976	0.5812	0.5893	0.9607
0.0084	9.0221	15500	0.2436	0.5966	0.5817	0.5891	0.9605
0.0066	9.3132	16000	0.2515	0.5982	0.5724	0.5850	0.9604
0.0075	9.6042	16500	0.2522	0.6046	0.5761	0.5900	0.9607
0.0076	9.8952	17000	0.2580	0.6131	0.5687	0.5901	0.9609
0.0055	10.1863	17500	0.2519	0.6093	0.5926	0.6008	0.9612
0.0053	10.4773	18000	0.2603	0.6078	0.5840	0.5957	0.9607

Framework versions

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