electra-small-discriminator-finetuned-ner-cadec

This model is a fine-tuned version of google/electra-small-discriminator on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3089
• Precision: 0.5240
• Recall: 0.6280
• F1: 0.5713
• Accuracy: 0.9120
• Adr Precision: 0.4580
• Adr Recall: 0.6495
• Adr F1: 0.5372
• Disease Precision: 0.0
• Disease Recall: 0.0
• Disease F1: 0.0
• Drug Precision: 0.8187
• Drug Recall: 0.9030
• Drug F1: 0.8588
• Finding Precision: 0.0
• Finding Recall: 0.0
• Finding F1: 0.0
• Symptom Precision: 0.0
• Symptom Recall: 0.0
• Symptom F1: 0.0
• B-adr Precision: 0.6345
• B-adr Recall: 0.8196
• B-adr F1: 0.7152
• B-disease Precision: 0.0
• B-disease Recall: 0.0
• B-disease F1: 0.0
• B-drug Precision: 0.8902
• B-drug Recall: 0.9333
• B-drug F1: 0.9112
• B-finding Precision: 0.0
• B-finding Recall: 0.0
• B-finding F1: 0.0
• B-symptom Precision: 0.0
• B-symptom Recall: 0.0
• B-symptom F1: 0.0
• I-adr Precision: 0.4753
• I-adr Recall: 0.6380
• I-adr F1: 0.5448
• I-disease Precision: 0.0
• I-disease Recall: 0.0
• I-disease F1: 0.0
• I-drug Precision: 0.8371
• I-drug Recall: 0.9030
• I-drug F1: 0.8688
• I-finding Precision: 0.6667
• I-finding Recall: 0.0625
• I-finding F1: 0.1143
• I-symptom Precision: 0.0
• I-symptom Recall: 0.0
• I-symptom F1: 0.0
• Macro Avg F1: 0.3154
• Weighted Avg F1: 0.6288

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• 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	Adr Precision	Adr Recall	Adr F1	Disease Precision	Disease Recall	Disease F1	Drug Precision	Drug Recall	Drug F1	Finding Precision	Finding Recall	Finding F1	Symptom Precision	Symptom Recall	Symptom F1	B-adr Precision	B-adr Recall	B-adr F1	B-disease Precision	B-disease Recall	B-disease F1	B-drug Precision	B-drug Recall	B-drug F1	B-finding Precision	B-finding Recall	B-finding F1	B-symptom Precision	B-symptom Recall	B-symptom F1	I-adr Precision	I-adr Recall	I-adr F1	I-disease Precision	I-disease Recall	I-disease F1	I-drug Precision	I-drug Recall	I-drug F1	I-finding Precision	I-finding Recall	I-finding F1	I-symptom Precision	I-symptom Recall	I-symptom F1	Macro Avg F1	Weighted Avg F1
No log	1.0	127	0.5920	0.2042	0.1710	0.1861	0.8334	0.2042	0.2514	0.2253	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1667	0.0038	0.0075	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0238	0.0353	0.0285	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0036	0.0115
No log	2.0	254	0.4272	0.4122	0.4806	0.4438	0.8895	0.3547	0.5376	0.4274	0.0	0.0	0.0	0.8519	0.5576	0.6740	0.0	0.0	0.0	0.0	0.0	0.0	0.5794	0.7351	0.6481	0.0	0.0	0.0	0.9709	0.6061	0.7463	0.0	0.0	0.0	0.0	0.0	0.0	0.3556	0.5055	0.4175	0.0	0.0	0.0	0.9588	0.5636	0.7099	0.0	0.0	0.0	0.0	0.0	0.0	0.2522	0.5261
No log	3.0	381	0.3799	0.4580	0.5918	0.5163	0.8982	0.4002	0.6294	0.4893	0.0	0.0	0.0	0.7360	0.7939	0.7638	0.0	0.0	0.0	0.0	0.0	0.0	0.5794	0.7985	0.6715	0.0	0.0	0.0	0.8758	0.8545	0.8650	0.0	0.0	0.0	0.0	0.0	0.0	0.4018	0.5960	0.4800	0.0	0.0	0.0	0.7870	0.8061	0.7964	0.0	0.0	0.0	0.0	0.0	0.0	0.2813	0.5771
0.6133	4.0	508	0.3578	0.4755	0.6067	0.5332	0.9007	0.4097	0.6330	0.4975	0.0	0.0	0.0	0.7833	0.8545	0.8174	0.0	0.0	0.0	0.0	0.0	0.0	0.5840	0.8273	0.6847	0.0	0.0	0.0	0.8862	0.8970	0.8916	0.0	0.0	0.0	0.0	0.0	0.0	0.4169	0.6093	0.4951	0.0	0.0	0.0	0.8372	0.8727	0.8546	0.0	0.0	0.0	0.0	0.0	0.0	0.2926	0.5961
0.6133	5.0	635	0.3254	0.5124	0.6180	0.5603	0.9079	0.4476	0.6422	0.5275	0.0	0.0	0.0	0.7880	0.8788	0.8309	0.0	0.0	0.0	0.0	0.0	0.0	0.6197	0.8100	0.7022	0.0	0.0	0.0	0.8613	0.9030	0.8817	0.0	0.0	0.0	0.0	0.0	0.0	0.4663	0.6269	0.5348	0.0	0.0	0.0	0.8066	0.8848	0.8439	0.0	0.0	0.0	0.0	0.0	0.0	0.2963	0.6124
0.6133	6.0	762	0.3153	0.5272	0.6180	0.5690	0.9114	0.4624	0.6422	0.5376	0.0	0.0	0.0	0.7967	0.8788	0.8357	0.0	0.0	0.0	0.0	0.0	0.0	0.6353	0.8023	0.7091	0.0	0.0	0.0	0.8779	0.9152	0.8961	0.0	0.0	0.0	0.0	0.0	0.0	0.4649	0.6137	0.5290	0.0	0.0	0.0	0.8249	0.8848	0.8538	0.0	0.0	0.0	0.0	0.0	0.0	0.2988	0.6158
0.6133	7.0	889	0.3095	0.5341	0.6155	0.5719	0.9126	0.4658	0.6367	0.5380	0.0	0.0	0.0	0.8202	0.8848	0.8513	0.0	0.0	0.0	0.0	0.0	0.0	0.6453	0.7927	0.7115	0.0	0.0	0.0	0.8935	0.9152	0.9042	0.0	0.0	0.0	0.0	0.0	0.0	0.4714	0.6181	0.5349	0.0	0.0	0.0	0.8448	0.8909	0.8673	0.0	0.0	0.0	0.0	0.0	0.0	0.3018	0.6209
0.3074	8.0	1016	0.3080	0.5292	0.6217	0.5718	0.9121	0.4604	0.6404	0.5357	0.0	0.0	0.0	0.8324	0.9030	0.8663	0.0	0.0	0.0	0.0	0.0	0.0	0.6397	0.8042	0.7126	0.0	0.0	0.0	0.8947	0.9273	0.9107	0.0	0.0	0.0	0.0	0.0	0.0	0.4792	0.6358	0.5465	0.0	0.0	0.0	0.8418	0.9030	0.8713	0.5	0.0312	0.0588	0.0	0.0	0.0	0.3100	0.6274
0.3074	9.0	1143	0.3072	0.5225	0.6230	0.5683	0.9119	0.4583	0.6459	0.5362	0.0	0.0	0.0	0.8077	0.8909	0.8473	0.0	0.0	0.0	0.0	0.0	0.0	0.6415	0.8138	0.7174	0.0	0.0	0.0	0.8882	0.9152	0.9015	0.0	0.0	0.0	0.0	0.0	0.0	0.4690	0.6336	0.5390	0.0	0.0	0.0	0.8362	0.8970	0.8655	0.3333	0.0312	0.0571	0.0	0.0	0.0	0.3081	0.6250
0.3074	10.0	1270	0.3089	0.5240	0.6280	0.5713	0.9120	0.4580	0.6495	0.5372	0.0	0.0	0.0	0.8187	0.9030	0.8588	0.0	0.0	0.0	0.0	0.0	0.0	0.6345	0.8196	0.7152	0.0	0.0	0.0	0.8902	0.9333	0.9112	0.0	0.0	0.0	0.0	0.0	0.0	0.4753	0.6380	0.5448	0.0	0.0	0.0	0.8371	0.9030	0.8688	0.6667	0.0625	0.1143	0.0	0.0	0.0	0.3154	0.6288

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0