fydhfzh/hubert-classifier-aug-fold-7-best

hubert-classifier-aug-fold-7

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: 0.4362
• Accuracy: 0.8976
• Precision: 0.9089
• Recall: 0.8976
• F1: 0.8953
• Binary: 0.9294

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	0.19	50	4.0533	0.0324	0.0014	0.0324	0.0027	0.2927
No log	0.38	100	3.5885	0.0514	0.0169	0.0514	0.0106	0.3262
No log	0.58	150	3.3501	0.0541	0.0206	0.0541	0.0140	0.3330
No log	0.77	200	3.2675	0.0649	0.0154	0.0649	0.0203	0.3405
No log	0.96	250	3.0965	0.1162	0.0330	0.1162	0.0456	0.3797
3.7373	1.15	300	2.9467	0.1514	0.0616	0.1514	0.0770	0.4035
3.7373	1.34	350	2.7676	0.2	0.1022	0.2	0.1111	0.4384
3.7373	1.53	400	2.5435	0.2649	0.1888	0.2649	0.1783	0.4805
3.7373	1.73	450	2.3812	0.2973	0.1993	0.2973	0.1965	0.5068
3.7373	1.92	500	2.1573	0.3946	0.3075	0.3946	0.3120	0.5749
2.9139	2.11	550	1.9561	0.4486	0.4077	0.4486	0.3798	0.6132
2.9139	2.3	600	1.7966	0.4784	0.4652	0.4784	0.4273	0.6324
2.9139	2.49	650	1.7610	0.5270	0.5083	0.5270	0.4675	0.6632
2.9139	2.68	700	1.5796	0.5351	0.4840	0.5351	0.4750	0.6741
2.9139	2.88	750	1.4707	0.5676	0.5624	0.5676	0.5239	0.6935
2.2164	3.07	800	1.3680	0.6162	0.6049	0.6162	0.5829	0.7286
2.2164	3.26	850	1.2484	0.6162	0.6078	0.6162	0.5800	0.7319
2.2164	3.45	900	1.1271	0.6649	0.6659	0.6649	0.6400	0.7646
2.2164	3.64	950	1.0343	0.7108	0.7139	0.7108	0.6851	0.7959
2.2164	3.84	1000	1.0379	0.7027	0.7076	0.7027	0.6733	0.7922
1.8319	4.03	1050	1.0744	0.7	0.7494	0.7	0.6818	0.7935
1.8319	4.22	1100	0.9615	0.7324	0.7692	0.7324	0.7194	0.8141
1.8319	4.41	1150	0.8683	0.7514	0.7827	0.7514	0.7341	0.8251
1.8319	4.6	1200	0.8870	0.7432	0.7827	0.7432	0.7307	0.8195
1.8319	4.79	1250	0.8191	0.7676	0.7874	0.7676	0.7516	0.8357
1.8319	4.99	1300	0.7923	0.7784	0.8235	0.7784	0.7701	0.8441
1.5844	5.18	1350	0.7525	0.8	0.8203	0.8	0.7905	0.8605
1.5844	5.37	1400	0.7352	0.8	0.8401	0.8	0.7994	0.8603
1.5844	5.56	1450	0.6931	0.8081	0.8423	0.8081	0.8017	0.8649
1.5844	5.75	1500	0.6872	0.8081	0.8367	0.8081	0.8005	0.8670
1.5844	5.94	1550	0.6630	0.8189	0.8507	0.8189	0.8133	0.8735
1.4042	6.14	1600	0.6284	0.8216	0.8424	0.8216	0.8153	0.8757
1.4042	6.33	1650	0.7190	0.7865	0.8274	0.7865	0.7788	0.8508
1.4042	6.52	1700	0.6470	0.8216	0.8428	0.8216	0.8163	0.8754
1.4042	6.71	1750	0.6415	0.8324	0.8655	0.8324	0.8277	0.8822
1.4042	6.9	1800	0.6644	0.8216	0.8554	0.8216	0.8133	0.8735
1.2826	7.09	1850	0.6328	0.8243	0.8607	0.8243	0.8217	0.8781
1.2826	7.29	1900	0.6106	0.8351	0.8673	0.8351	0.8284	0.8857
1.2826	7.48	1950	0.6186	0.8297	0.8686	0.8297	0.8248	0.8803
1.2826	7.67	2000	0.6167	0.8351	0.8709	0.8351	0.8321	0.8838
1.2826	7.86	2050	0.5680	0.8378	0.8691	0.8378	0.8352	0.8857
1.1959	8.05	2100	0.5415	0.8541	0.8849	0.8541	0.8512	0.8978
1.1959	8.25	2150	0.5322	0.8568	0.8910	0.8568	0.8552	0.8997
1.1959	8.44	2200	0.5865	0.8432	0.8675	0.8432	0.8373	0.8914
1.1959	8.63	2250	0.5779	0.8541	0.8865	0.8541	0.8512	0.9000
1.1959	8.82	2300	0.5011	0.8757	0.9080	0.8757	0.8752	0.9154
1.1236	9.01	2350	0.5108	0.8514	0.8804	0.8514	0.8498	0.8981
1.1236	9.2	2400	0.5375	0.8486	0.8772	0.8486	0.8459	0.8962
1.1236	9.4	2450	0.5775	0.8459	0.8746	0.8459	0.8473	0.8943
1.1236	9.59	2500	0.5318	0.8514	0.8862	0.8514	0.8497	0.9003
1.1236	9.78	2550	0.5484	0.8459	0.8761	0.8459	0.8439	0.8976
1.1236	9.97	2600	0.5733	0.8486	0.8849	0.8486	0.8474	0.8951
1.0544	10.16	2650	0.5349	0.8541	0.8818	0.8541	0.8496	0.9000
1.0544	10.35	2700	0.5435	0.8459	0.8777	0.8459	0.8388	0.8932
1.0544	10.55	2750	0.4787	0.8595	0.8822	0.8595	0.8563	0.9027
1.0544	10.74	2800	0.4678	0.8595	0.8880	0.8595	0.8562	0.9027
1.0544	10.93	2850	0.4572	0.8730	0.9001	0.8730	0.8707	0.9103
1.0171	11.12	2900	0.5138	0.8568	0.8876	0.8568	0.8529	0.8997
1.0171	11.31	2950	0.5102	0.8757	0.8980	0.8757	0.8750	0.9130
1.0171	11.51	3000	0.5265	0.8676	0.8921	0.8676	0.8649	0.9076
1.0171	11.7	3050	0.4659	0.8730	0.8961	0.8730	0.8733	0.9132
1.0171	11.89	3100	0.4995	0.8676	0.8917	0.8676	0.8621	0.9084
0.9541	12.08	3150	0.4533	0.8811	0.8996	0.8811	0.8788	0.9168
0.9541	12.27	3200	0.4571	0.8865	0.9085	0.8865	0.8866	0.9205
0.9541	12.46	3250	0.4846	0.8622	0.8908	0.8622	0.8596	0.9035
0.9541	12.66	3300	0.4850	0.8730	0.8989	0.8730	0.8710	0.9111
0.9541	12.85	3350	0.4826	0.8568	0.8834	0.8568	0.8522	0.8997
0.9149	13.04	3400	0.4680	0.8730	0.8938	0.8730	0.8717	0.9103
0.9149	13.23	3450	0.5733	0.8486	0.8769	0.8486	0.8468	0.8941
0.9149	13.42	3500	0.5068	0.8730	0.8975	0.8730	0.8718	0.9111
0.9149	13.61	3550	0.4816	0.8730	0.8991	0.8730	0.8721	0.9122
0.9149	13.81	3600	0.5007	0.8676	0.8944	0.8676	0.8677	0.9095
0.9149	14.0	3650	0.4674	0.8811	0.9061	0.8811	0.8796	0.9168
0.8802	14.19	3700	0.4997	0.8622	0.8860	0.8622	0.8608	0.9035
0.8802	14.38	3750	0.4425	0.8784	0.9036	0.8784	0.8768	0.9149
0.8802	14.57	3800	0.5111	0.8811	0.9088	0.8811	0.8808	0.9170
0.8802	14.77	3850	0.4408	0.8811	0.9036	0.8811	0.8794	0.9168
0.8802	14.96	3900	0.5053	0.8622	0.8855	0.8622	0.8570	0.9035
0.8475	15.15	3950	0.5046	0.8622	0.8897	0.8622	0.8599	0.9038
0.8475	15.34	4000	0.4560	0.8649	0.8849	0.8649	0.8635	0.9068
0.8475	15.53	4050	0.4562	0.8730	0.8944	0.8730	0.8722	0.9124
0.8475	15.72	4100	0.4827	0.8622	0.8932	0.8622	0.8611	0.9027
0.8475	15.92	4150	0.4750	0.8784	0.9039	0.8784	0.8775	0.9159
0.8235	16.11	4200	0.4789	0.8703	0.8998	0.8703	0.8689	0.9092
0.8235	16.3	4250	0.4445	0.8892	0.9136	0.8892	0.8875	0.9227
0.8235	16.49	4300	0.4804	0.8703	0.8950	0.8703	0.8690	0.9086
0.8235	16.68	4350	0.4556	0.8676	0.8878	0.8676	0.8639	0.9076
0.8235	16.87	4400	0.5254	0.8622	0.8844	0.8622	0.8571	0.9030
0.7913	17.07	4450	0.4432	0.8946	0.9105	0.8946	0.8916	0.9273
0.7913	17.26	4500	0.4991	0.8622	0.8906	0.8622	0.8603	0.9035
0.7913	17.45	4550	0.4480	0.8865	0.9067	0.8865	0.8836	0.9205
0.7913	17.64	4600	0.4408	0.8757	0.8954	0.8757	0.8748	0.9130
0.7913	17.83	4650	0.4559	0.8811	0.9033	0.8811	0.8804	0.9189
0.7769	18.02	4700	0.4716	0.8919	0.9136	0.8919	0.8914	0.9254
0.7769	18.22	4750	0.4492	0.8811	0.9059	0.8811	0.8785	0.9170
0.7769	18.41	4800	0.4714	0.8811	0.9062	0.8811	0.8798	0.9170
0.7769	18.6	4850	0.4849	0.8757	0.9015	0.8757	0.8745	0.9122
0.7769	18.79	4900	0.4156	0.8946	0.9140	0.8946	0.8918	0.9262
0.7769	18.98	4950	0.4333	0.8892	0.9066	0.8892	0.8862	0.9227
0.7461	19.18	5000	0.4054	0.9054	0.9220	0.9054	0.9033	0.9341
0.7461	19.37	5050	0.4613	0.8757	0.8999	0.8757	0.8699	0.9132
0.7461	19.56	5100	0.4379	0.8865	0.9112	0.8865	0.8854	0.9219
0.7461	19.75	5150	0.4349	0.8946	0.9120	0.8946	0.8934	0.9262
0.7461	19.94	5200	0.4647	0.8811	0.9009	0.8811	0.8794	0.9181
0.7216	20.13	5250	0.4346	0.9027	0.9189	0.9027	0.9017	0.9322
0.7216	20.33	5300	0.4577	0.9	0.9156	0.9	0.8984	0.9322
0.7216	20.52	5350	0.4712	0.8946	0.9152	0.8946	0.8944	0.9276

Framework versions

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