hubert-classifier-aug-80

This model is a fine-tuned version of facebook/hubert-base-ls960 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7879
• Accuracy: 0.8248
• Precision: 0.8454
• Recall: 0.8248
• F1: 0.8211
• Binary: 0.8778

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Binary
No log	0.22	50	4.4263	0.0135	0.0002	0.0135	0.0004	0.1245
No log	0.43	100	4.4241	0.0148	0.0002	0.0148	0.0004	0.1261
No log	0.65	150	4.4226	0.0135	0.0002	0.0135	0.0004	0.1333
No log	0.86	200	4.0724	0.0310	0.0015	0.0310	0.0027	0.2334
No log	1.08	250	3.9071	0.0270	0.0009	0.0270	0.0016	0.2287
No log	1.29	300	3.8704	0.0283	0.0009	0.0283	0.0017	0.2327
No log	1.51	350	3.8439	0.0256	0.0007	0.0256	0.0014	0.2306
No log	1.73	400	3.8189	0.0256	0.0015	0.0256	0.0025	0.2330
No log	1.94	450	3.7439	0.0391	0.0058	0.0391	0.0068	0.2950
4.1116	2.16	500	3.5560	0.0404	0.0033	0.0404	0.0046	0.3185
4.1116	2.37	550	3.4895	0.0404	0.0035	0.0404	0.0059	0.3150
4.1116	2.59	600	3.3966	0.0512	0.0053	0.0512	0.0094	0.3253
4.1116	2.8	650	3.3257	0.0687	0.0157	0.0687	0.0175	0.3420
4.1116	3.02	700	3.2077	0.0593	0.0106	0.0593	0.0156	0.3136
4.1116	3.24	750	3.1384	0.0957	0.0225	0.0957	0.0302	0.3609
4.1116	3.45	800	3.1351	0.0943	0.0273	0.0943	0.0364	0.3597
4.1116	3.67	850	2.9107	0.1253	0.0606	0.1253	0.0646	0.3829
4.1116	3.88	900	2.8937	0.1509	0.0677	0.1509	0.0712	0.3927
4.1116	4.1	950	2.7737	0.1806	0.1052	0.1806	0.1128	0.4205
3.2639	4.31	1000	2.6901	0.1658	0.0774	0.1658	0.0851	0.4117
3.2639	4.53	1050	2.5441	0.2183	0.1329	0.2183	0.1424	0.4482
3.2639	4.75	1100	2.4408	0.2197	0.1407	0.2197	0.1382	0.4488
3.2639	4.96	1150	2.4113	0.2278	0.1691	0.2278	0.1517	0.4562
3.2639	5.18	1200	2.2525	0.2790	0.2052	0.2790	0.1916	0.4904
3.2639	5.39	1250	2.2126	0.2817	0.2064	0.2817	0.1939	0.4962
3.2639	5.61	1300	2.1644	0.2951	0.2583	0.2951	0.2264	0.5039
3.2639	5.83	1350	2.1951	0.3275	0.2823	0.3275	0.2698	0.5199
3.2639	6.04	1400	1.9989	0.3666	0.3230	0.3666	0.3087	0.5532
3.2639	6.26	1450	1.8910	0.3962	0.3735	0.3962	0.3340	0.5749
2.4809	6.47	1500	1.8342	0.4084	0.4092	0.4084	0.3542	0.5838
2.4809	6.69	1550	1.8166	0.4272	0.4345	0.4272	0.3809	0.5954
2.4809	6.9	1600	1.6498	0.4838	0.4594	0.4838	0.4297	0.6345
2.4809	7.12	1650	1.6093	0.5040	0.5262	0.5040	0.4666	0.6515
2.4809	7.34	1700	1.5510	0.5296	0.5257	0.5296	0.4896	0.6689
2.4809	7.55	1750	1.5003	0.5175	0.5164	0.5175	0.4669	0.6621
2.4809	7.77	1800	1.4597	0.5270	0.5263	0.5270	0.4861	0.6671
2.4809	7.98	1850	1.3801	0.5916	0.6024	0.5916	0.5598	0.7115
2.4809	8.2	1900	1.3262	0.5863	0.5970	0.5863	0.5574	0.7101
2.4809	8.41	1950	1.2342	0.5943	0.5938	0.5943	0.5648	0.7163
1.8737	8.63	2000	1.2114	0.6173	0.6210	0.6173	0.5957	0.7333
1.8737	8.85	2050	1.1831	0.6321	0.6535	0.6321	0.6072	0.7414
1.8737	9.06	2100	1.1501	0.6563	0.6882	0.6563	0.6398	0.7567
1.8737	9.28	2150	1.0732	0.6941	0.7109	0.6941	0.6802	0.7841
1.8737	9.49	2200	1.1194	0.6604	0.6696	0.6604	0.6424	0.7615
1.8737	9.71	2250	0.9827	0.7035	0.7331	0.7035	0.6924	0.7926
1.8737	9.92	2300	0.9956	0.7156	0.7381	0.7156	0.7047	0.8007
1.8737	10.14	2350	1.0312	0.6698	0.7095	0.6698	0.6552	0.7685
1.8737	10.36	2400	0.9753	0.7197	0.7465	0.7197	0.7070	0.8043
1.8737	10.57	2450	0.9825	0.7237	0.7322	0.7237	0.7118	0.8074
1.4378	10.79	2500	0.9829	0.6927	0.7234	0.6927	0.6821	0.7853
1.4378	11.0	2550	0.8897	0.7251	0.7515	0.7251	0.7152	0.8066
1.4378	11.22	2600	0.8627	0.7345	0.7624	0.7345	0.7277	0.8123
1.4378	11.43	2650	0.8772	0.7264	0.7602	0.7264	0.7228	0.8074
1.4378	11.65	2700	0.9209	0.7399	0.7622	0.7399	0.7321	0.8164
1.4378	11.87	2750	0.8737	0.7412	0.7623	0.7412	0.7345	0.8181
1.4378	12.08	2800	0.8638	0.7439	0.7632	0.7439	0.7370	0.8189
1.4378	12.3	2850	0.8525	0.7547	0.7763	0.7547	0.7492	0.8290
1.4378	12.51	2900	0.8238	0.7466	0.7598	0.7466	0.7382	0.8209
1.4378	12.73	2950	0.8192	0.7507	0.7771	0.7507	0.7446	0.8241
1.1771	12.94	3000	0.7660	0.7642	0.7801	0.7642	0.7589	0.8338
1.1771	13.16	3050	0.8528	0.7453	0.7676	0.7453	0.7369	0.8213
1.1771	13.38	3100	0.7580	0.7776	0.7881	0.7776	0.7707	0.8425
1.1771	13.59	3150	0.8186	0.7615	0.7849	0.7615	0.7536	0.8345
1.1771	13.81	3200	0.7512	0.7871	0.8057	0.7871	0.7808	0.8519
1.1771	14.02	3250	0.7426	0.7763	0.7965	0.7763	0.7710	0.8439
1.1771	14.24	3300	0.8203	0.7695	0.7827	0.7695	0.7619	0.8407
1.1771	14.46	3350	0.7871	0.7682	0.7878	0.7682	0.7590	0.8377
1.1771	14.67	3400	0.7761	0.7830	0.8044	0.7830	0.7733	0.8470
1.1771	14.89	3450	0.8547	0.7763	0.7965	0.7763	0.7731	0.8451
0.9984	15.1	3500	0.7879	0.7709	0.7922	0.7709	0.7633	0.8400
0.9984	15.32	3550	0.7582	0.8086	0.8235	0.8086	0.8037	0.8655
0.9984	15.53	3600	0.7084	0.7938	0.8074	0.7938	0.7872	0.8555
0.9984	15.75	3650	0.7424	0.7911	0.8099	0.7911	0.7864	0.8553
0.9984	15.97	3700	0.7255	0.8127	0.8274	0.8127	0.8090	0.8706
0.9984	16.18	3750	0.6903	0.8059	0.8216	0.8059	0.8011	0.8646
0.9984	16.4	3800	0.7078	0.8100	0.8324	0.8100	0.8043	0.8689
0.9984	16.61	3850	0.7843	0.7992	0.8218	0.7992	0.7940	0.8604
0.9984	16.83	3900	0.7239	0.7965	0.8226	0.7965	0.7936	0.8581
0.9984	17.04	3950	0.7097	0.8127	0.8283	0.8127	0.8092	0.8679
0.8969	17.26	4000	0.8020	0.7951	0.8135	0.7951	0.7914	0.8566
0.8969	17.48	4050	0.6915	0.8275	0.8477	0.8275	0.8242	0.8792
0.8969	17.69	4100	0.7548	0.8113	0.8321	0.8113	0.8071	0.8685
0.8969	17.91	4150	0.7284	0.8073	0.8293	0.8073	0.8036	0.8673
0.8969	18.12	4200	0.7304	0.8127	0.8276	0.8127	0.8092	0.8687
0.8969	18.34	4250	0.7169	0.8154	0.8319	0.8154	0.8109	0.8706
0.8969	18.55	4300	0.7189	0.8194	0.8375	0.8194	0.8173	0.8736
0.8969	18.77	4350	0.8506	0.7790	0.8073	0.7790	0.7718	0.8457
0.8969	18.99	4400	0.7322	0.8248	0.8436	0.8248	0.8211	0.8788
0.8969	19.2	4450	0.7497	0.8032	0.8212	0.8032	0.7997	0.8627
0.8076	19.42	4500	0.7879	0.8248	0.8454	0.8248	0.8211	0.8778
0.8076	19.63	4550	0.8195	0.8073	0.8314	0.8073	0.8032	0.8660
0.8076	19.85	4600	0.8176	0.8059	0.8249	0.8059	0.8032	0.8651
0.8076	20.06	4650	0.7699	0.8221	0.8375	0.8221	0.8180	0.8753
0.8076	20.28	4700	0.7316	0.8181	0.8407	0.8181	0.8162	0.8722
0.8076	20.5	4750	0.7205	0.8086	0.8272	0.8086	0.8060	0.8670
0.8076	20.71	4800	0.7689	0.7965	0.8150	0.7965	0.7930	0.8574
0.8076	20.93	4850	0.7828	0.8127	0.8297	0.8127	0.8101	0.8689

Framework versions

• Transformers 4.38.2
• Pytorch 2.3.0
• Datasets 2.19.1
• Tokenizers 0.15.1