fydhfzh/hubert-classifier-aug-fold-1-old

hubert-classifier-aug-fold-1

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: 0.5129
• Accuracy: 0.8706
• Precision: 0.8830
• Recall: 0.8706
• F1: 0.8687
• Binary: 0.9084

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	3.8853	0.0499	0.0125	0.0499	0.0132	0.3132
No log	0.43	100	3.4193	0.0702	0.0103	0.0702	0.0167	0.3417
No log	0.65	150	3.1606	0.1161	0.0528	0.1161	0.0512	0.3783
No log	0.86	200	2.9650	0.1714	0.1093	0.1714	0.1008	0.4148
3.6314	1.08	250	2.7976	0.1714	0.0947	0.1714	0.0927	0.4148
3.6314	1.29	300	2.4874	0.2740	0.1723	0.2740	0.1770	0.4870
3.6314	1.51	350	2.3192	0.3414	0.2741	0.3414	0.2588	0.5379
3.6314	1.72	400	2.1116	0.4278	0.3772	0.4278	0.3514	0.5988
3.6314	1.94	450	1.9495	0.4683	0.4712	0.4683	0.4075	0.6273
2.6994	2.16	500	1.8632	0.4764	0.4337	0.4764	0.4098	0.6313
2.6994	2.37	550	1.6822	0.5250	0.4922	0.5250	0.4621	0.6664
2.6994	2.59	600	1.5299	0.5938	0.5590	0.5938	0.5296	0.7165
2.6994	2.8	650	1.4177	0.6140	0.5885	0.6140	0.5664	0.7298
2.1575	3.02	700	1.3614	0.6100	0.5558	0.6100	0.5498	0.7283
2.1575	3.23	750	1.2209	0.6491	0.6650	0.6491	0.6086	0.7528
2.1575	3.45	800	1.2059	0.6829	0.7281	0.6829	0.6553	0.7748
2.1575	3.66	850	1.1207	0.7031	0.7301	0.7031	0.6741	0.7908
2.1575	3.88	900	1.0553	0.7139	0.7368	0.7139	0.6967	0.7993
1.8321	4.09	950	0.9864	0.7490	0.7412	0.7490	0.7291	0.8250
1.8321	4.31	1000	0.9365	0.7544	0.7653	0.7544	0.7360	0.8278
1.8321	4.53	1050	0.9050	0.7584	0.7617	0.7584	0.7438	0.8323
1.8321	4.74	1100	0.8241	0.7760	0.7896	0.7760	0.7638	0.8424
1.8321	4.96	1150	0.8260	0.7611	0.7712	0.7611	0.7455	0.8335
1.6136	5.17	1200	0.8008	0.7814	0.8013	0.7814	0.7738	0.8474
1.6136	5.39	1250	0.7588	0.7787	0.7981	0.7787	0.7694	0.8453
1.6136	5.6	1300	0.7619	0.7868	0.7963	0.7868	0.7755	0.8520
1.6136	5.82	1350	0.7611	0.7868	0.7972	0.7868	0.7755	0.8514
1.4406	6.03	1400	0.7094	0.7976	0.8090	0.7976	0.7923	0.8591
1.4406	6.25	1450	0.7133	0.7976	0.7990	0.7976	0.7873	0.8595
1.4406	6.47	1500	0.7126	0.8016	0.8151	0.8016	0.7921	0.8618
1.4406	6.68	1550	0.6381	0.8043	0.8242	0.8043	0.7979	0.8642
1.4406	6.9	1600	0.6248	0.8111	0.8253	0.8111	0.8079	0.8683
1.3033	7.11	1650	0.6333	0.8138	0.8391	0.8138	0.8125	0.8699
1.3033	7.33	1700	0.6087	0.8232	0.8450	0.8232	0.8188	0.8775
1.3033	7.54	1750	0.5919	0.8273	0.8347	0.8273	0.8199	0.8796
1.3033	7.76	1800	0.5835	0.8232	0.8398	0.8232	0.8198	0.8779
1.3033	7.97	1850	0.5697	0.8232	0.8397	0.8232	0.8192	0.8780
1.2408	8.19	1900	0.6186	0.8111	0.8354	0.8111	0.8050	0.8694
1.2408	8.41	1950	0.5622	0.8327	0.8445	0.8327	0.8306	0.8830
1.2408	8.62	2000	0.5740	0.8205	0.8407	0.8205	0.8184	0.8756
1.2408	8.84	2050	0.5627	0.8300	0.8460	0.8300	0.8286	0.8816
1.1272	9.05	2100	0.5340	0.8381	0.8538	0.8381	0.8354	0.8873
1.1272	9.27	2150	0.5927	0.8246	0.8475	0.8246	0.8205	0.8795
1.1272	9.48	2200	0.5429	0.8300	0.8511	0.8300	0.8292	0.8812
1.1272	9.7	2250	0.4781	0.8529	0.8659	0.8529	0.8518	0.8977
1.1272	9.91	2300	0.5081	0.8448	0.8621	0.8448	0.8442	0.8930
1.0783	10.13	2350	0.5015	0.8489	0.8680	0.8489	0.8473	0.8947
1.0783	10.34	2400	0.5225	0.8421	0.8627	0.8421	0.8389	0.8906
1.0783	10.56	2450	0.5113	0.8435	0.8602	0.8435	0.8426	0.8911
1.0783	10.78	2500	0.5009	0.8556	0.8712	0.8556	0.8526	0.9005
1.0783	10.99	2550	0.5112	0.8421	0.8571	0.8421	0.8416	0.8922
1.0384	11.21	2600	0.4812	0.8543	0.8663	0.8543	0.8516	0.8996
1.0384	11.42	2650	0.5017	0.8583	0.8731	0.8583	0.8561	0.9020
1.0384	11.64	2700	0.4837	0.8489	0.8650	0.8489	0.8450	0.8958
1.0384	11.85	2750	0.5084	0.8435	0.8643	0.8435	0.8389	0.8920
1.0115	12.07	2800	0.4843	0.8583	0.8767	0.8583	0.8562	0.9024
1.0115	12.28	2850	0.5396	0.8502	0.8703	0.8502	0.8465	0.8973
1.0115	12.5	2900	0.5440	0.8435	0.8563	0.8435	0.8375	0.8916
1.0115	12.72	2950	0.5076	0.8489	0.8600	0.8489	0.8460	0.8958
1.0115	12.93	3000	0.5253	0.8435	0.8571	0.8435	0.8422	0.8915
0.9461	13.15	3050	0.4841	0.8664	0.8828	0.8664	0.8669	0.9081
0.9461	13.36	3100	0.4890	0.8637	0.8759	0.8637	0.8626	0.9062
0.9461	13.58	3150	0.4948	0.8664	0.8819	0.8664	0.8656	0.9076
0.9461	13.79	3200	0.4851	0.8596	0.8763	0.8596	0.8590	0.9028
0.9001	14.01	3250	0.4966	0.8583	0.8735	0.8583	0.8587	0.9020
0.9001	14.22	3300	0.5335	0.8543	0.8688	0.8543	0.8531	0.8985
0.9001	14.44	3350	0.5072	0.8596	0.8732	0.8596	0.8587	0.9024
0.9001	14.66	3400	0.4968	0.8664	0.8793	0.8664	0.8654	0.9076
0.9001	14.87	3450	0.4763	0.8637	0.8779	0.8637	0.8629	0.9057
0.8663	15.09	3500	0.5134	0.8704	0.8781	0.8704	0.8685	0.9104
0.8663	15.3	3550	0.5010	0.8664	0.8812	0.8664	0.8668	0.9081
0.8663	15.52	3600	0.5221	0.8556	0.8694	0.8556	0.8547	0.9000
0.8663	15.73	3650	0.4990	0.8704	0.8813	0.8704	0.8691	0.9104
0.8663	15.95	3700	0.5046	0.8610	0.8729	0.8610	0.8591	0.9032
0.8499	16.16	3750	0.5099	0.8583	0.8730	0.8583	0.8562	0.9019
0.8499	16.38	3800	0.4859	0.8677	0.8786	0.8677	0.8668	0.9085
0.8499	16.59	3850	0.5005	0.8637	0.8731	0.8637	0.8619	0.9067
0.8499	16.81	3900	0.5169	0.8596	0.8735	0.8596	0.8590	0.9023
0.8149	17.03	3950	0.5035	0.8583	0.8710	0.8583	0.8579	0.9019
0.8149	17.24	4000	0.5211	0.8570	0.8748	0.8570	0.8573	0.9009

Framework versions

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