fydhfzh/hubert-classifier-aug-fold-2-old

hubert-classifier-aug-fold-2

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: 0.5511
• Accuracy: 0.8720
• Precision: 0.8843
• Recall: 0.8720
• F1: 0.8715
• Binary: 0.9111

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	4.3908	0.0364	0.0036	0.0364	0.0055	0.2323
No log	0.43	100	3.7968	0.0418	0.0037	0.0418	0.0062	0.3105
No log	0.65	150	3.4223	0.0499	0.0034	0.0499	0.0062	0.3297
No log	0.86	200	3.2649	0.0985	0.0263	0.0985	0.0345	0.3641
3.8781	1.08	250	3.1085	0.1323	0.0462	0.1323	0.0530	0.3862
3.8781	1.29	300	3.0185	0.1228	0.0541	0.1228	0.0555	0.3795
3.8781	1.51	350	2.8481	0.1660	0.1305	0.1660	0.1055	0.4117
3.8781	1.72	400	2.6281	0.2254	0.1349	0.2254	0.1446	0.4561
3.8781	1.94	450	2.4175	0.2605	0.1934	0.2605	0.1835	0.4804
3.0628	2.16	500	2.2785	0.2915	0.1990	0.2915	0.1961	0.5027
3.0628	2.37	550	2.0868	0.3738	0.3460	0.3738	0.3058	0.5586
3.0628	2.59	600	1.9295	0.4305	0.3815	0.4305	0.3534	0.6031
3.0628	2.8	650	1.7881	0.4872	0.4541	0.4872	0.4192	0.6405
2.478	3.02	700	1.6403	0.5304	0.4907	0.5304	0.4651	0.6715
2.478	3.23	750	1.5843	0.5735	0.5809	0.5735	0.5262	0.7026
2.478	3.45	800	1.4142	0.5789	0.5783	0.5789	0.5349	0.7040
2.478	3.66	850	1.2960	0.6289	0.6269	0.6289	0.5828	0.7399
2.478	3.88	900	1.2609	0.6680	0.6867	0.6680	0.6347	0.7686
2.0301	4.09	950	1.1055	0.6991	0.7211	0.6991	0.6727	0.7906
2.0301	4.31	1000	1.1323	0.7018	0.7338	0.7018	0.6907	0.7918
2.0301	4.53	1050	0.9934	0.7314	0.7459	0.7314	0.7152	0.8112
2.0301	4.74	1100	0.9489	0.7395	0.7599	0.7395	0.7213	0.8174
2.0301	4.96	1150	0.9370	0.7355	0.7633	0.7355	0.7260	0.8151
1.7314	5.17	1200	0.8762	0.7679	0.7919	0.7679	0.7548	0.8359
1.7314	5.39	1250	0.7944	0.7868	0.8047	0.7868	0.7783	0.8520
1.7314	5.6	1300	0.7927	0.7881	0.8091	0.7881	0.7804	0.8518
1.7314	5.82	1350	0.7881	0.7895	0.8087	0.7895	0.7830	0.8533
1.5038	6.03	1400	0.7181	0.8057	0.8260	0.8057	0.8012	0.8661
1.5038	6.25	1450	0.7339	0.8084	0.8205	0.8084	0.8006	0.8665
1.5038	6.47	1500	0.6845	0.8192	0.8371	0.8192	0.8160	0.8737
1.5038	6.68	1550	0.6878	0.8246	0.8463	0.8246	0.8239	0.8776
1.5038	6.9	1600	0.6396	0.8300	0.8439	0.8300	0.8277	0.8803
1.3718	7.11	1650	0.6743	0.8232	0.8458	0.8232	0.8225	0.8772
1.3718	7.33	1700	0.6018	0.8435	0.8644	0.8435	0.8419	0.8912
1.3718	7.54	1750	0.6461	0.8300	0.8494	0.8300	0.8293	0.8811
1.3718	7.76	1800	0.6181	0.8178	0.8385	0.8178	0.8157	0.8731
1.3718	7.97	1850	0.5696	0.8421	0.8608	0.8421	0.8425	0.8904
1.2723	8.19	1900	0.5924	0.8354	0.8550	0.8354	0.8335	0.8858
1.2723	8.41	1950	0.6176	0.8313	0.8464	0.8313	0.8281	0.8821
1.2723	8.62	2000	0.6529	0.8246	0.8496	0.8246	0.8239	0.8773
1.2723	8.84	2050	0.5532	0.8489	0.8680	0.8489	0.8485	0.8953
1.1706	9.05	2100	0.5791	0.8502	0.8673	0.8502	0.8502	0.8957
1.1706	9.27	2150	0.5871	0.8448	0.8624	0.8448	0.8434	0.8916
1.1706	9.48	2200	0.5853	0.8394	0.8605	0.8394	0.8394	0.8887
1.1706	9.7	2250	0.5420	0.8502	0.8675	0.8502	0.8494	0.8962
1.1706	9.91	2300	0.5913	0.8286	0.8515	0.8286	0.8275	0.8822
1.102	10.13	2350	0.5674	0.8448	0.8614	0.8448	0.8426	0.8927
1.102	10.34	2400	0.5948	0.8502	0.8675	0.8502	0.8491	0.8966
1.102	10.56	2450	0.5462	0.8543	0.8725	0.8543	0.8544	0.8969
1.102	10.78	2500	0.5774	0.8462	0.8662	0.8462	0.8469	0.8928
1.102	10.99	2550	0.5601	0.8448	0.8671	0.8448	0.8460	0.8938
1.0635	11.21	2600	0.5655	0.8570	0.8744	0.8570	0.8572	0.8988
1.0635	11.42	2650	0.5635	0.8516	0.8688	0.8516	0.8506	0.8950
1.0635	11.64	2700	0.6053	0.8327	0.8547	0.8327	0.8325	0.8822
1.0635	11.85	2750	0.6079	0.8408	0.8621	0.8408	0.8390	0.8896
1.0079	12.07	2800	0.5631	0.8489	0.8657	0.8489	0.8489	0.8962
1.0079	12.28	2850	0.5383	0.8691	0.8828	0.8691	0.8685	0.9093
1.0079	12.5	2900	0.5289	0.8623	0.8760	0.8623	0.8620	0.9040
1.0079	12.72	2950	0.5366	0.8650	0.8841	0.8650	0.8661	0.9065
1.0079	12.93	3000	0.5507	0.8516	0.8715	0.8516	0.8507	0.8965
0.9768	13.15	3050	0.5615	0.8596	0.8734	0.8596	0.8586	0.9036
0.9768	13.36	3100	0.5332	0.8583	0.8769	0.8583	0.8591	0.9016
0.9768	13.58	3150	0.5017	0.8691	0.8825	0.8691	0.8687	0.9086
0.9768	13.79	3200	0.5430	0.8623	0.8753	0.8623	0.8619	0.9034
0.9196	14.01	3250	0.5612	0.8556	0.8706	0.8556	0.8536	0.8992
0.9196	14.22	3300	0.5152	0.8664	0.8817	0.8664	0.8652	0.9062
0.9196	14.44	3350	0.5082	0.8745	0.8852	0.8745	0.8744	0.9126
0.9196	14.66	3400	0.5131	0.8745	0.8824	0.8745	0.8744	0.9128
0.9196	14.87	3450	0.5532	0.8543	0.8691	0.8543	0.8529	0.8992
0.8897	15.09	3500	0.5611	0.8556	0.8682	0.8556	0.8544	0.8987
0.8897	15.3	3550	0.5185	0.8664	0.8742	0.8664	0.8646	0.9062
0.8897	15.52	3600	0.5576	0.8556	0.8713	0.8556	0.8547	0.8988
0.8897	15.73	3650	0.5190	0.8650	0.8758	0.8650	0.8631	0.9053
0.8897	15.95	3700	0.5215	0.8623	0.8753	0.8623	0.8613	0.9031
0.8483	16.16	3750	0.5424	0.8543	0.8708	0.8543	0.8524	0.8978
0.8483	16.38	3800	0.5499	0.8691	0.8814	0.8691	0.8698	0.9090
0.8483	16.59	3850	0.5676	0.8556	0.8711	0.8556	0.8530	0.8997
0.8483	16.81	3900	0.5817	0.8570	0.8750	0.8570	0.8565	0.8996
0.8359	17.03	3950	0.5602	0.8583	0.8759	0.8583	0.8576	0.9005
0.8359	17.24	4000	0.5563	0.8596	0.8760	0.8596	0.8589	0.9016
0.8359	17.46	4050	0.5181	0.8637	0.8754	0.8637	0.8620	0.9039
0.8359	17.67	4100	0.5184	0.8731	0.8886	0.8731	0.8710	0.9119
0.8359	17.89	4150	0.5222	0.8664	0.8755	0.8664	0.8636	0.9076
0.7893	18.1	4200	0.5228	0.8691	0.8800	0.8691	0.8683	0.9086
0.7893	18.32	4250	0.5159	0.8745	0.8879	0.8745	0.8743	0.9128
0.7893	18.53	4300	0.5181	0.8745	0.8843	0.8745	0.8724	0.9117
0.7893	18.75	4350	0.5315	0.8772	0.8884	0.8772	0.8759	0.9134
0.7893	18.97	4400	0.5018	0.8772	0.8880	0.8772	0.8756	0.9132
0.768	19.18	4450	0.5367	0.8677	0.8787	0.8677	0.8682	0.9067
0.768	19.4	4500	0.5166	0.8745	0.8855	0.8745	0.8733	0.9120
0.768	19.61	4550	0.5010	0.8758	0.8850	0.8758	0.8748	0.9128
0.768	19.83	4600	0.5553	0.8704	0.8813	0.8704	0.8694	0.9096
0.7463	20.04	4650	0.5474	0.8637	0.8755	0.8637	0.8625	0.9053
0.7463	20.26	4700	0.5482	0.8664	0.8795	0.8664	0.8661	0.9062
0.7463	20.47	4750	0.5528	0.8650	0.8782	0.8650	0.8649	0.9058
0.7463	20.69	4800	0.5490	0.8650	0.8782	0.8650	0.8640	0.9053
0.7463	20.91	4850	0.5362	0.8637	0.8746	0.8637	0.8620	0.9043
0.7261	21.12	4900	0.5296	0.8691	0.8830	0.8691	0.8681	0.9077

Framework versions

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