fydhfzh/hubert-classifier-aug-fold-0-old

hubert-classifier-aug-fold-0

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: 0.5592
• Accuracy: 0.8464
• Precision: 0.8588
• Recall: 0.8464
• F1: 0.8431
• Binary: 0.8926

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: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Binary
No log	0.22	50	4.4295	0.0135	0.0002	0.0135	0.0004	0.1332
No log	0.43	100	4.4254	0.0148	0.0002	0.0148	0.0004	0.1274
No log	0.65	150	3.8186	0.0364	0.0121	0.0364	0.0050	0.3090
No log	0.86	200	3.5321	0.0391	0.0090	0.0391	0.0062	0.3193
4.1413	1.08	250	3.3337	0.0728	0.0256	0.0728	0.0286	0.3453
4.1413	1.29	300	3.1664	0.0970	0.0489	0.0970	0.0400	0.3590
4.1413	1.51	350	2.9961	0.1253	0.0613	0.1253	0.0631	0.3821
4.1413	1.73	400	2.8225	0.1739	0.0798	0.1739	0.0904	0.4181
4.1413	1.94	450	2.6439	0.2116	0.1109	0.2116	0.1236	0.4457
3.2276	2.16	500	2.4578	0.2385	0.1802	0.2385	0.1570	0.4670
3.2276	2.37	550	2.2801	0.3396	0.2831	0.3396	0.2516	0.5358
3.2276	2.59	600	2.0684	0.4003	0.3030	0.4003	0.3068	0.5796
3.2276	2.8	650	1.9308	0.4299	0.3493	0.4299	0.3516	0.6005
2.5852	3.02	700	1.8448	0.4501	0.4000	0.4501	0.3811	0.6146
2.5852	3.24	750	1.6568	0.5283	0.4743	0.5283	0.4552	0.6689
2.5852	3.45	800	1.6974	0.4690	0.4551	0.4690	0.4169	0.6264
2.5852	3.67	850	1.4828	0.5687	0.5769	0.5687	0.5231	0.6978
2.5852	3.88	900	1.4420	0.5580	0.5477	0.5580	0.5126	0.6896
2.1226	4.1	950	1.3306	0.6186	0.6133	0.6186	0.5784	0.7315
2.1226	4.31	1000	1.2209	0.6456	0.6561	0.6456	0.6076	0.7500
2.1226	4.53	1050	1.1256	0.6698	0.6865	0.6698	0.6404	0.7664
2.1226	4.75	1100	1.0700	0.6846	0.7003	0.6846	0.6586	0.7770
2.1226	4.96	1150	1.0085	0.7156	0.7415	0.7156	0.6942	0.7993
1.8257	5.18	1200	1.0190	0.7224	0.7397	0.7224	0.7028	0.8046
1.8257	5.39	1250	0.9742	0.7102	0.7244	0.7102	0.6886	0.7961
1.8257	5.61	1300	0.8793	0.7561	0.7680	0.7561	0.7384	0.8284
1.8257	5.83	1350	0.8472	0.7547	0.7763	0.7547	0.7426	0.8280
1.5842	6.04	1400	0.8424	0.7601	0.7956	0.7601	0.7487	0.8327
1.5842	6.26	1450	0.7802	0.7642	0.7846	0.7642	0.7513	0.8348
1.5842	6.47	1500	0.7447	0.7965	0.8096	0.7965	0.7914	0.8574
1.5842	6.69	1550	0.7081	0.7844	0.8035	0.7844	0.7772	0.8499
1.5842	6.9	1600	0.7616	0.7722	0.7995	0.7722	0.7681	0.8399
1.4387	7.12	1650	0.7133	0.7709	0.7904	0.7709	0.7607	0.8403
1.4387	7.34	1700	0.6570	0.8127	0.8301	0.8127	0.8094	0.8695
1.4387	7.55	1750	0.6325	0.8221	0.8461	0.8221	0.8212	0.8761
1.4387	7.77	1800	0.6352	0.8032	0.8251	0.8032	0.8004	0.8625
1.4387	7.98	1850	0.6313	0.8086	0.8270	0.8086	0.8040	0.8678
1.3174	8.2	1900	0.6843	0.8154	0.8372	0.8154	0.8100	0.8710
1.3174	8.41	1950	0.6142	0.8194	0.8360	0.8194	0.8153	0.8739
1.3174	8.63	2000	0.6324	0.8154	0.8229	0.8154	0.8102	0.8710
1.3174	8.85	2050	0.5751	0.8383	0.8566	0.8383	0.8351	0.8852
1.2131	9.06	2100	0.5873	0.8275	0.8439	0.8275	0.8250	0.8805
1.2131	9.28	2150	0.6016	0.8167	0.8346	0.8167	0.8131	0.8729
1.2131	9.49	2200	0.5982	0.8410	0.8617	0.8410	0.8387	0.8879
1.2131	9.71	2250	0.5490	0.8437	0.8564	0.8437	0.8410	0.8912
1.2131	9.92	2300	0.5587	0.8342	0.8537	0.8342	0.8309	0.8837
1.1426	10.14	2350	0.5969	0.8261	0.8446	0.8261	0.8214	0.8790
1.1426	10.36	2400	0.5936	0.8410	0.8575	0.8410	0.8382	0.8889
1.1426	10.57	2450	0.5656	0.8383	0.8579	0.8383	0.8364	0.8865
1.1426	10.79	2500	0.5130	0.8625	0.8756	0.8625	0.8593	0.9054
1.0738	11.0	2550	0.5832	0.8396	0.8618	0.8396	0.8389	0.8880
1.0738	11.22	2600	0.5554	0.8423	0.8634	0.8423	0.8417	0.8908
1.0738	11.43	2650	0.5763	0.8275	0.8490	0.8275	0.8238	0.8801
1.0738	11.65	2700	0.5697	0.8329	0.8452	0.8329	0.8281	0.8857
1.0738	11.87	2750	0.5413	0.8464	0.8655	0.8464	0.8432	0.8922
1.0326	12.08	2800	0.5954	0.8235	0.8443	0.8235	0.8176	0.8761
1.0326	12.3	2850	0.5665	0.8410	0.8611	0.8410	0.8354	0.8908

Framework versions

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