electra-base-discriminator-finetuned-ner-cadec-no-iob

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

• Loss: 0.3907
• Precision: 0.6459
• Recall: 0.6715
• F1: 0.6585
• Accuracy: 0.9355
• Adr Precision: 0.5961
• Adr Recall: 0.6330
• Adr F1: 0.614
• Disease Precision: 0.4615
• Disease Recall: 0.375
• Disease F1: 0.4138
• Drug Precision: 0.9022
• Drug Recall: 0.9222
• Drug F1: 0.9121
• Finding Precision: 0.2571
• Finding Recall: 0.2812
• Finding F1: 0.2687
• Symptom Precision: 0.5357
• Symptom Recall: 0.5172
• Symptom F1: 0.5263
• Macro Avg F1: 0.5470
• Weighted Avg F1: 0.6584

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: 40

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	Macro Avg F1	Weighted Avg F1
No log	1.0	125	0.2178	0.5120	0.5646	0.5370	0.9209	0.4311	0.5608	0.4875	0.1	0.0312	0.0476	0.8158	0.8611	0.8378	0.0	0.0	0.0	0.0	0.0	0.0	0.2746	0.5129
No log	2.0	250	0.1864	0.5799	0.6227	0.6005	0.9296	0.5017	0.6	0.5465	0.4146	0.5312	0.4658	0.9011	0.9111	0.9061	0.0	0.0	0.0	0.0	0.0	0.0	0.3837	0.5845
No log	3.0	375	0.1845	0.5894	0.6306	0.6093	0.9310	0.5520	0.6021	0.5759	0.2353	0.375	0.2892	0.9016	0.9167	0.9091	0.0857	0.0938	0.0896	0.4615	0.2069	0.2857	0.4299	0.6113
0.2159	4.0	500	0.1968	0.6002	0.6359	0.6176	0.9318	0.5570	0.6041	0.5796	0.0667	0.0312	0.0426	0.9121	0.9222	0.9171	0.1967	0.375	0.2581	0.5263	0.3448	0.4167	0.4428	0.6173
0.2159	5.0	625	0.2026	0.6357	0.6583	0.6468	0.9338	0.5817	0.6165	0.5986	0.52	0.4062	0.4561	0.9071	0.9222	0.9146	0.1951	0.25	0.2192	0.5909	0.4483	0.5098	0.5397	0.6482
0.2159	6.0	750	0.2142	0.6055	0.6359	0.6203	0.9319	0.5697	0.5814	0.5755	0.375	0.4688	0.4167	0.9066	0.9167	0.9116	0.1163	0.1562	0.1333	0.4167	0.5172	0.4615	0.4997	0.6256
0.2159	7.0	875	0.2283	0.6420	0.6649	0.6533	0.9328	0.5867	0.6206	0.6032	0.5	0.3125	0.3846	0.9076	0.9278	0.9176	0.2444	0.3438	0.2857	0.6522	0.5172	0.5769	0.5536	0.6542
0.0651	8.0	1000	0.2430	0.6103	0.6570	0.6328	0.9312	0.5632	0.6062	0.5839	0.5625	0.2812	0.375	0.9121	0.9222	0.9171	0.2241	0.4062	0.2889	0.4211	0.5517	0.4776	0.5285	0.6377
0.0651	9.0	1125	0.2367	0.6308	0.6583	0.6443	0.9338	0.5800	0.6206	0.5996	0.48	0.375	0.4211	0.8811	0.9056	0.8932	0.2286	0.25	0.2388	0.5556	0.5172	0.5357	0.5377	0.6441
0.0651	10.0	1250	0.2694	0.6241	0.6570	0.6401	0.9294	0.5639	0.6186	0.5900	0.4286	0.2812	0.3396	0.9171	0.9222	0.9197	0.2222	0.3125	0.2597	0.6842	0.4483	0.5417	0.5301	0.6419
0.0651	11.0	1375	0.2731	0.61	0.6438	0.6264	0.9325	0.5626	0.6021	0.5817	0.4286	0.375	0.4000	0.8511	0.8889	0.8696	0.2812	0.2812	0.2812	0.4545	0.5172	0.4839	0.5233	0.6259
0.0265	12.0	1500	0.2842	0.6221	0.6451	0.6334	0.9323	0.5697	0.5979	0.5835	0.3929	0.3438	0.3667	0.9027	0.9278	0.9151	0.2121	0.2188	0.2154	0.4516	0.4828	0.4667	0.5095	0.6331
0.0265	13.0	1625	0.2861	0.6438	0.6675	0.6554	0.9348	0.5802	0.6268	0.6026	0.5	0.4375	0.4667	0.9278	0.9278	0.9278	0.3	0.2812	0.2903	0.5	0.4138	0.4528	0.5480	0.6552
0.0265	14.0	1750	0.3202	0.6413	0.6557	0.6484	0.9323	0.5742	0.6144	0.5936	0.5263	0.3125	0.3922	0.9071	0.9222	0.9146	0.2647	0.2812	0.2727	0.7	0.4828	0.5714	0.5489	0.6469
0.0265	15.0	1875	0.3118	0.6284	0.6491	0.6385	0.9336	0.5731	0.6062	0.5892	0.4348	0.3125	0.3636	0.9066	0.9167	0.9116	0.25	0.3125	0.2778	0.52	0.4483	0.4815	0.5247	0.6390
0.0115	16.0	2000	0.3144	0.6293	0.6583	0.6435	0.9335	0.5747	0.6268	0.5996	0.3846	0.3125	0.3448	0.9061	0.9111	0.9086	0.2286	0.25	0.2388	0.5909	0.4483	0.5098	0.5203	0.6436
0.0115	17.0	2125	0.3235	0.6411	0.6623	0.6515	0.9346	0.5918	0.6247	0.6078	0.4	0.375	0.3871	0.9171	0.9222	0.9197	0.2121	0.2188	0.2154	0.5185	0.4828	0.5	0.5260	0.6519
0.0115	18.0	2250	0.3258	0.6230	0.6583	0.6402	0.9344	0.5948	0.6144	0.6045	0.3429	0.375	0.3582	0.8913	0.9111	0.9011	0.1957	0.2812	0.2308	0.4571	0.5517	0.5	0.5189	0.6447
0.0115	19.0	2375	0.3347	0.6312	0.6728	0.6513	0.9361	0.5794	0.6392	0.6078	0.4348	0.3125	0.3636	0.9022	0.9222	0.9121	0.2564	0.3125	0.2817	0.5185	0.4828	0.5	0.5331	0.6519
0.0076	20.0	2500	0.3334	0.6508	0.6662	0.6584	0.9328	0.6031	0.6330	0.6177	0.375	0.375	0.375	0.9231	0.9333	0.9282	0.1765	0.1875	0.1818	0.6316	0.4138	0.5	0.5205	0.6583
0.0076	21.0	2625	0.3426	0.6385	0.6781	0.6577	0.9344	0.5943	0.6495	0.6207	0.4231	0.3438	0.3793	0.9071	0.9222	0.9146	0.1765	0.1875	0.1818	0.5	0.5517	0.5246	0.5242	0.6581
0.0076	22.0	2750	0.3431	0.6269	0.6649	0.6453	0.9344	0.5862	0.6309	0.6077	0.3793	0.3438	0.3607	0.9071	0.9222	0.9146	0.1707	0.2188	0.1918	0.4828	0.4828	0.4828	0.5115	0.6478
0.0076	23.0	2875	0.3374	0.6410	0.6689	0.6546	0.9348	0.5913	0.6412	0.6152	0.3913	0.2812	0.3273	0.9180	0.9333	0.9256	0.2368	0.2812	0.2571	0.4762	0.3448	0.4000	0.5051	0.6534
0.0049	24.0	3000	0.3484	0.6338	0.6530	0.6433	0.9352	0.5819	0.6082	0.5948	0.4	0.375	0.3871	0.9126	0.9278	0.9201	0.2258	0.2188	0.2222	0.4667	0.4828	0.4746	0.5198	0.6429
0.0049	25.0	3125	0.3441	0.6444	0.6741	0.6589	0.9362	0.6015	0.6412	0.6208	0.4194	0.4062	0.4127	0.9076	0.9278	0.9176	0.1935	0.1875	0.1905	0.4667	0.4828	0.4746	0.5232	0.6587
0.0049	26.0	3250	0.3573	0.6330	0.6689	0.6504	0.9347	0.5802	0.6268	0.6026	0.4138	0.375	0.3934	0.9235	0.9389	0.9311	0.2286	0.25	0.2388	0.4667	0.4828	0.4746	0.5281	0.6515
0.0049	27.0	3375	0.3589	0.6477	0.6596	0.6536	0.9336	0.5909	0.6165	0.6034	0.48	0.375	0.4211	0.9126	0.9278	0.9201	0.25	0.2812	0.2647	0.5909	0.4483	0.5098	0.5438	0.6531
0.0037	28.0	3500	0.3629	0.6307	0.6715	0.6505	0.9354	0.5900	0.6351	0.6117	0.4444	0.375	0.4068	0.9016	0.9167	0.9091	0.2326	0.3125	0.2667	0.4375	0.4828	0.4590	0.5307	0.6533
0.0037	29.0	3625	0.3694	0.6490	0.6781	0.6632	0.9358	0.6035	0.6433	0.6228	0.4286	0.375	0.4000	0.9071	0.9222	0.9146	0.2105	0.25	0.2286	0.6154	0.5517	0.5818	0.5495	0.6644
0.0037	30.0	3750	0.3779	0.6456	0.6728	0.6589	0.9350	0.5946	0.6351	0.6142	0.5	0.375	0.4286	0.9071	0.9222	0.9146	0.2647	0.2812	0.2727	0.4839	0.5172	0.5000	0.5460	0.6589
0.0037	31.0	3875	0.3776	0.6465	0.6636	0.6549	0.9360	0.5992	0.6227	0.6107	0.4615	0.375	0.4138	0.9022	0.9222	0.9121	0.2424	0.25	0.2462	0.4839	0.5172	0.5000	0.5366	0.6543
0.0027	32.0	4000	0.3775	0.6429	0.6675	0.6550	0.9355	0.5984	0.6268	0.6123	0.4444	0.375	0.4068	0.9022	0.9222	0.9121	0.2368	0.2812	0.2571	0.5	0.5172	0.5085	0.5394	0.6558
0.0027	33.0	4125	0.3784	0.6378	0.6715	0.6542	0.9363	0.5881	0.6330	0.6097	0.4783	0.3438	0.4	0.9022	0.9222	0.9121	0.2381	0.3125	0.2703	0.5556	0.5172	0.5357	0.5456	0.6555
0.0027	34.0	4250	0.3824	0.6332	0.6741	0.6530	0.9356	0.5830	0.6371	0.6089	0.4286	0.375	0.4000	0.9022	0.9222	0.9121	0.2432	0.2812	0.2609	0.5357	0.5172	0.5263	0.5416	0.6542
0.0027	35.0	4375	0.3825	0.6415	0.6728	0.6568	0.9358	0.5954	0.6371	0.6155	0.4444	0.375	0.4068	0.9022	0.9222	0.9121	0.225	0.2812	0.25	0.56	0.4828	0.5185	0.5406	0.6580
0.0018	36.0	4500	0.3841	0.6390	0.6794	0.6586	0.9364	0.5928	0.6454	0.6180	0.4444	0.375	0.4068	0.9022	0.9222	0.9121	0.2368	0.2812	0.2571	0.5172	0.5172	0.5172	0.5422	0.6598
0.0018	37.0	4625	0.3892	0.6477	0.6741	0.6606	0.9361	0.6008	0.6392	0.6194	0.4286	0.375	0.4000	0.9022	0.9222	0.9121	0.2571	0.2812	0.2687	0.5385	0.4828	0.5091	0.5418	0.6606
0.0018	38.0	4750	0.3893	0.6480	0.6702	0.6589	0.9363	0.5988	0.6309	0.6145	0.4444	0.375	0.4068	0.9022	0.9222	0.9121	0.25	0.2812	0.2647	0.5769	0.5172	0.5455	0.5487	0.6590
0.0018	39.0	4875	0.3911	0.6422	0.6702	0.6559	0.9354	0.5907	0.6309	0.6102	0.4615	0.375	0.4138	0.9022	0.9222	0.9121	0.2571	0.2812	0.2687	0.5357	0.5172	0.5263	0.5462	0.6559
0.0016	40.0	5000	0.3907	0.6459	0.6715	0.6585	0.9355	0.5961	0.6330	0.614	0.4615	0.375	0.4138	0.9022	0.9222	0.9121	0.2571	0.2812	0.2687	0.5357	0.5172	0.5263	0.5470	0.6584

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0