hubert-classifier-aug-large

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: 0.5737
• Accuracy: 0.8518
• Precision: 0.8710
• Recall: 0.8518
• F1: 0.8459
• Binary: 0.8946

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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Binary
No log	0.1	50	4.2254	0.0296	0.0051	0.0296	0.0050	0.2472
No log	0.2	100	3.6044	0.0499	0.0034	0.0499	0.0062	0.3265
No log	0.29	150	3.4158	0.0620	0.0137	0.0620	0.0182	0.3314
No log	0.39	200	3.1970	0.0943	0.0301	0.0943	0.0402	0.3632
No log	0.49	250	3.1359	0.1253	0.0438	0.1253	0.0550	0.3786
No log	0.59	300	2.8477	0.1941	0.0877	0.1941	0.1046	0.4311
No log	0.69	350	2.5968	0.2291	0.1477	0.2291	0.1464	0.4569
No log	0.78	400	2.3869	0.3019	0.2233	0.3019	0.1981	0.5092
No log	0.88	450	2.2318	0.3342	0.2618	0.3342	0.2491	0.5337
3.3246	0.98	500	2.0726	0.4057	0.3384	0.4057	0.3220	0.5819
3.3246	1.08	550	1.9390	0.4043	0.3492	0.4043	0.3267	0.5813
3.3246	1.18	600	1.8723	0.4394	0.3484	0.4394	0.3548	0.6063
3.3246	1.27	650	1.7220	0.5081	0.4868	0.5081	0.4462	0.6571
3.3246	1.37	700	1.5947	0.5283	0.4654	0.5283	0.4610	0.6691
3.3246	1.47	750	1.5081	0.5512	0.5536	0.5512	0.5010	0.6863
3.3246	1.57	800	1.3927	0.6078	0.6098	0.6078	0.5698	0.7252
3.3246	1.67	850	1.2970	0.6361	0.6405	0.6361	0.6036	0.7445
3.3246	1.76	900	1.2218	0.6658	0.6832	0.6658	0.6470	0.7663
3.3246	1.86	950	1.2574	0.6725	0.7113	0.6725	0.6538	0.7698
2.1116	1.96	1000	1.0768	0.7224	0.7288	0.7224	0.7073	0.8058
2.1116	2.06	1050	1.0574	0.7318	0.7445	0.7318	0.7160	0.8113
2.1116	2.16	1100	0.9994	0.7332	0.7526	0.7332	0.7171	0.8139
2.1116	2.25	1150	0.9494	0.7358	0.7497	0.7358	0.7196	0.8159
2.1116	2.35	1200	0.8719	0.7588	0.7743	0.7588	0.7456	0.8306
2.1116	2.45	1250	0.8674	0.7642	0.7862	0.7642	0.7530	0.8345
2.1116	2.55	1300	0.8805	0.7857	0.8075	0.7857	0.7754	0.8487
2.1116	2.65	1350	0.8389	0.7682	0.7955	0.7682	0.7601	0.8377
2.1116	2.75	1400	0.8189	0.7763	0.7913	0.7763	0.7640	0.8411
2.1116	2.84	1450	0.7739	0.7871	0.7881	0.7871	0.7737	0.8491
1.5744	2.94	1500	0.7971	0.7668	0.7887	0.7668	0.7567	0.8360
1.5744	3.04	1550	0.7348	0.7844	0.7998	0.7844	0.7781	0.8478
1.5744	3.14	1600	0.7241	0.7925	0.8115	0.7925	0.7838	0.8534
1.5744	3.24	1650	0.7763	0.7749	0.8010	0.7749	0.7709	0.8411
1.5744	3.33	1700	0.6638	0.8073	0.8278	0.8073	0.8036	0.8646
1.5744	3.43	1750	0.7065	0.8100	0.8339	0.8100	0.8021	0.8656
1.5744	3.53	1800	0.7391	0.7951	0.8212	0.7951	0.7878	0.8561
1.5744	3.63	1850	0.6450	0.8181	0.8385	0.8181	0.8125	0.8721
1.5744	3.73	1900	0.6834	0.8113	0.8342	0.8113	0.8076	0.8670
1.5744	3.82	1950	0.6616	0.8113	0.8311	0.8113	0.8025	0.8668
1.312	3.92	2000	0.6177	0.8194	0.8388	0.8194	0.8149	0.8718
1.312	4.02	2050	0.6550	0.8059	0.8338	0.8059	0.7979	0.8637
1.312	4.12	2100	0.5995	0.8261	0.8438	0.8261	0.8206	0.8772
1.312	4.22	2150	0.6833	0.8127	0.8340	0.8127	0.8069	0.8678
1.312	4.31	2200	0.6185	0.8275	0.8438	0.8275	0.8232	0.8787
1.312	4.41	2250	0.6334	0.8167	0.8364	0.8167	0.8111	0.8708
1.312	4.51	2300	0.6100	0.8208	0.8372	0.8208	0.8171	0.8725
1.312	4.61	2350	0.5953	0.8302	0.8477	0.8302	0.8268	0.8805
1.312	4.71	2400	0.5847	0.8194	0.8278	0.8194	0.8125	0.8735
1.312	4.8	2450	0.5932	0.8329	0.8475	0.8329	0.8295	0.8814
1.1711	4.9	2500	0.5890	0.8329	0.8511	0.8329	0.8285	0.8811
1.1711	5.0	2550	0.5662	0.8437	0.8574	0.8437	0.8374	0.8885
1.1711	5.1	2600	0.5648	0.8531	0.8657	0.8531	0.8494	0.8953
1.1711	5.2	2650	0.5805	0.8410	0.8555	0.8410	0.8385	0.8858
1.1711	5.29	2700	0.5372	0.8477	0.8631	0.8477	0.8426	0.8919
1.1711	5.39	2750	0.5698	0.8491	0.8661	0.8491	0.8481	0.8919
1.1711	5.49	2800	0.5499	0.8571	0.8756	0.8571	0.8525	0.8993
1.1711	5.59	2850	0.5643	0.8504	0.8671	0.8504	0.8477	0.8941
1.1711	5.69	2900	0.5834	0.8491	0.8649	0.8491	0.8461	0.8923
1.1711	5.78	2950	0.5306	0.8612	0.8760	0.8612	0.8580	0.9004
1.078	5.88	3000	0.5276	0.8571	0.8738	0.8571	0.8532	0.8980
1.078	5.98	3050	0.5294	0.8531	0.8792	0.8531	0.8507	0.8960
1.078	6.08	3100	0.5305	0.8558	0.8740	0.8558	0.8525	0.8974
1.078	6.18	3150	0.5546	0.8518	0.8700	0.8518	0.8488	0.8946
1.078	6.27	3200	0.5292	0.8652	0.8844	0.8652	0.8613	0.9049
1.078	6.37	3250	0.5871	0.8288	0.8527	0.8288	0.8270	0.8796
1.078	6.47	3300	0.5549	0.8477	0.8656	0.8477	0.8438	0.8914
1.078	6.57	3350	0.5559	0.8410	0.8544	0.8410	0.8349	0.8876
1.078	6.67	3400	0.5496	0.8531	0.8688	0.8531	0.8496	0.8961
1.078	6.76	3450	0.5737	0.8518	0.8710	0.8518	0.8459	0.8946

Framework versions

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