fydhfzh/hubert-classifier-aug-fold-4-old

hubert-classifier-aug-fold-4

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: 0.6062
• Accuracy: 0.8491
• Precision: 0.8645
• Recall: 0.8491
• F1: 0.8497
• Binary: 0.8941

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	4.3309	0.0297	0.0020	0.0297	0.0036	0.2159
No log	0.43	100	3.6722	0.0553	0.0262	0.0553	0.0174	0.3324
No log	0.65	150	3.3950	0.0621	0.0161	0.0621	0.0170	0.3379
No log	0.86	200	3.2145	0.1134	0.0428	0.1134	0.0506	0.3713
3.8422	1.08	250	3.0635	0.1525	0.0902	0.1525	0.0873	0.4015
3.8422	1.29	300	2.8389	0.2119	0.1084	0.2119	0.1229	0.4354
3.8422	1.51	350	2.5852	0.2348	0.1770	0.2348	0.1484	0.4611
3.8422	1.72	400	2.3922	0.2888	0.2381	0.2888	0.2034	0.4977
3.8422	1.94	450	2.2068	0.3185	0.2658	0.3185	0.2420	0.5202
2.9357	2.16	500	2.0744	0.4103	0.4153	0.4103	0.3548	0.5827
2.9357	2.37	550	1.8778	0.4588	0.3960	0.4588	0.3847	0.6197
2.9357	2.59	600	1.8338	0.4440	0.4075	0.4440	0.3748	0.6085
2.9357	2.8	650	1.6534	0.4777	0.4265	0.4777	0.4116	0.6337
2.3301	3.02	700	1.5716	0.5344	0.5094	0.5344	0.4768	0.6733
2.3301	3.23	750	1.4864	0.5655	0.5298	0.5655	0.5208	0.6926
2.3301	3.45	800	1.4863	0.5452	0.5615	0.5452	0.5023	0.6803
2.3301	3.66	850	1.3882	0.5655	0.5432	0.5655	0.5064	0.6949
2.3301	3.88	900	1.3190	0.5897	0.5922	0.5897	0.5473	0.7117
1.9679	4.09	950	1.1769	0.6559	0.6830	0.6559	0.6272	0.7580
1.9679	4.31	1000	1.2232	0.6289	0.6469	0.6289	0.6012	0.7402
1.9679	4.53	1050	1.1078	0.6694	0.6994	0.6694	0.6583	0.7675
1.9679	4.74	1100	1.0104	0.7072	0.7350	0.7072	0.6881	0.7949
1.9679	4.96	1150	0.9920	0.7152	0.7264	0.7152	0.6941	0.8016
1.7302	5.17	1200	0.9389	0.7449	0.7648	0.7449	0.7335	0.8208
1.7302	5.39	1250	0.8678	0.7611	0.7735	0.7611	0.7461	0.8344
1.7302	5.6	1300	0.8991	0.7449	0.7474	0.7449	0.7261	0.8217
1.7302	5.82	1350	0.7903	0.7665	0.7756	0.7665	0.7544	0.8368
1.531	6.03	1400	0.8221	0.7665	0.7852	0.7665	0.7528	0.8355
1.531	6.25	1450	0.7516	0.7773	0.8021	0.7773	0.7606	0.8436
1.531	6.47	1500	0.7573	0.7787	0.7870	0.7787	0.7655	0.8452
1.531	6.68	1550	0.7561	0.7719	0.7814	0.7719	0.7595	0.8401
1.531	6.9	1600	0.6898	0.8030	0.8144	0.8030	0.7957	0.8609
1.3639	7.11	1650	0.6912	0.7989	0.8155	0.7989	0.7916	0.8603
1.3639	7.33	1700	0.6771	0.7989	0.8201	0.7989	0.7915	0.8590
1.3639	7.54	1750	0.6652	0.8016	0.8207	0.8016	0.7952	0.8617
1.3639	7.76	1800	0.7130	0.8016	0.8227	0.8016	0.7972	0.8613
1.3639	7.97	1850	0.6661	0.7962	0.8173	0.7962	0.7878	0.8575
1.2774	8.19	1900	0.6532	0.8178	0.8313	0.8178	0.8130	0.8722
1.2774	8.41	1950	0.6291	0.8178	0.8307	0.8178	0.8144	0.8729
1.2774	8.62	2000	0.6585	0.8030	0.8183	0.8030	0.7958	0.8618
1.2774	8.84	2050	0.6427	0.8084	0.8216	0.8084	0.8038	0.8656
1.1807	9.05	2100	0.7025	0.8124	0.8246	0.8124	0.8081	0.8686
1.1807	9.27	2150	0.6977	0.8016	0.8211	0.8016	0.7980	0.8599
1.1807	9.48	2200	0.6439	0.8246	0.8404	0.8246	0.8223	0.8760
1.1807	9.7	2250	0.5890	0.8394	0.8503	0.8394	0.8374	0.8860
1.1807	9.91	2300	0.5406	0.8313	0.8429	0.8313	0.8281	0.8816
1.105	10.13	2350	0.6131	0.8232	0.8371	0.8232	0.8203	0.8742
1.105	10.34	2400	0.6241	0.8232	0.8421	0.8232	0.8206	0.8738
1.105	10.56	2450	0.6349	0.8354	0.8498	0.8354	0.8328	0.8837
1.105	10.78	2500	0.7053	0.8165	0.8290	0.8165	0.8118	0.8713
1.105	10.99	2550	0.5652	0.8381	0.8504	0.8381	0.8353	0.8866
1.0741	11.21	2600	0.5764	0.8408	0.8533	0.8408	0.8378	0.8887
1.0741	11.42	2650	0.5663	0.8448	0.8612	0.8448	0.8441	0.8915
1.0741	11.64	2700	0.6290	0.8219	0.8361	0.8219	0.8194	0.8741
1.0741	11.85	2750	0.5994	0.8381	0.8546	0.8381	0.8355	0.8873
1.0208	12.07	2800	0.5851	0.8327	0.8434	0.8327	0.8289	0.8826
1.0208	12.28	2850	0.6522	0.8219	0.8411	0.8219	0.8204	0.8746
1.0208	12.5	2900	0.6401	0.8273	0.8392	0.8273	0.8241	0.8779
1.0208	12.72	2950	0.5764	0.8475	0.8607	0.8475	0.8448	0.8930
1.0208	12.93	3000	0.5834	0.8354	0.8432	0.8354	0.8315	0.8839
0.9784	13.15	3050	0.6171	0.8394	0.8562	0.8394	0.8367	0.8872
0.9784	13.36	3100	0.6362	0.8300	0.8428	0.8300	0.8258	0.8798
0.9784	13.58	3150	0.6154	0.8313	0.8466	0.8313	0.8301	0.8807
0.9784	13.79	3200	0.5939	0.8421	0.8561	0.8421	0.8395	0.8887
0.9237	14.01	3250	0.6167	0.8435	0.8516	0.8435	0.8412	0.8900
0.9237	14.22	3300	0.6338	0.8408	0.8490	0.8408	0.8388	0.8887
0.9237	14.44	3350	0.6051	0.8421	0.8558	0.8421	0.8408	0.8900
0.9237	14.66	3400	0.5816	0.8367	0.8493	0.8367	0.8343	0.8854
0.9237	14.87	3450	0.6617	0.8327	0.8512	0.8327	0.8309	0.8825
0.8932	15.09	3500	0.6038	0.8448	0.8590	0.8448	0.8439	0.8915
0.8932	15.3	3550	0.6460	0.8408	0.8543	0.8408	0.8389	0.8883
0.8932	15.52	3600	0.5571	0.8489	0.8586	0.8489	0.8474	0.8943
0.8932	15.73	3650	0.6321	0.8273	0.8420	0.8273	0.8251	0.8792
0.8932	15.95	3700	0.6127	0.8448	0.8598	0.8448	0.8422	0.8919
0.856	16.16	3750	0.5622	0.8502	0.8620	0.8502	0.8492	0.8957
0.856	16.38	3800	0.5919	0.8529	0.8648	0.8529	0.8508	0.8981
0.856	16.59	3850	0.5345	0.8489	0.8582	0.8489	0.8468	0.8947
0.856	16.81	3900	0.6384	0.8408	0.8585	0.8408	0.8399	0.8883
0.8435	17.03	3950	0.5643	0.8596	0.8724	0.8596	0.8582	0.9024
0.8435	17.24	4000	0.5582	0.8448	0.8573	0.8448	0.8441	0.8911
0.8435	17.46	4050	0.5755	0.8489	0.8590	0.8489	0.8467	0.8943
0.8435	17.67	4100	0.6104	0.8340	0.8478	0.8340	0.8309	0.8839
0.8435	17.89	4150	0.6327	0.8435	0.8568	0.8435	0.8413	0.8906
0.7849	18.1	4200	0.5686	0.8583	0.8732	0.8583	0.8566	0.9009
0.7849	18.32	4250	0.6191	0.8462	0.8622	0.8462	0.8430	0.8924
0.7849	18.53	4300	0.5812	0.8489	0.8641	0.8489	0.8491	0.8934
0.7849	18.75	4350	0.5785	0.8529	0.8678	0.8529	0.8519	0.8966
0.7849	18.97	4400	0.5474	0.8502	0.8638	0.8502	0.8484	0.8957
0.779	19.18	4450	0.5515	0.8637	0.8753	0.8637	0.8620	0.9047
0.779	19.4	4500	0.5591	0.8623	0.8736	0.8623	0.8610	0.9047
0.779	19.61	4550	0.5287	0.8718	0.8806	0.8718	0.8700	0.9108
0.779	19.83	4600	0.5561	0.8610	0.8704	0.8610	0.8598	0.9032
0.7593	20.04	4650	0.5487	0.8583	0.8686	0.8583	0.8566	0.9013
0.7593	20.26	4700	0.5873	0.8502	0.8649	0.8502	0.8470	0.8957
0.7593	20.47	4750	0.5470	0.8475	0.8603	0.8475	0.8454	0.8934
0.7593	20.69	4800	0.5921	0.8543	0.8680	0.8543	0.8521	0.8977
0.7593	20.91	4850	0.5692	0.8543	0.8658	0.8543	0.8523	0.8981
0.7392	21.12	4900	0.5863	0.8394	0.8527	0.8394	0.8370	0.8881

Framework versions

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