hubert-classifier-aug-fold-3

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: 0.6214
• Accuracy: 0.8544
• Precision: 0.8720
• Recall: 0.8544
• F1: 0.8540
• Binary: 0.8989

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
• lr_scheduler_warmup_steps: 500
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Binary
No log	0.24	50	4.4176	0.0172	0.0140	0.0172	0.0083	0.1549
No log	0.48	100	4.3072	0.0434	0.0497	0.0434	0.0197	0.2778
No log	0.72	150	3.9604	0.0906	0.0606	0.0906	0.0446	0.3581
No log	0.96	200	3.6355	0.1191	0.0507	0.1191	0.0558	0.3783
4.235	1.2	250	3.3489	0.1700	0.0800	0.1700	0.0858	0.4157
4.235	1.44	300	3.1008	0.2015	0.1270	0.2015	0.1158	0.4382
4.235	1.68	350	2.8220	0.2906	0.2075	0.2906	0.2012	0.5014
4.235	1.92	400	2.5557	0.3843	0.3160	0.3843	0.3099	0.5671
3.2055	2.16	450	2.1790	0.4801	0.4047	0.4801	0.4036	0.6344
3.2055	2.4	500	1.9034	0.5790	0.5557	0.5790	0.5261	0.7028
3.2055	2.63	550	1.6707	0.6135	0.6116	0.6135	0.5701	0.7273
3.2055	2.87	600	1.4658	0.6285	0.6047	0.6285	0.5817	0.7381
2.1878	3.11	650	1.3665	0.6457	0.6522	0.6457	0.6153	0.7534
2.1878	3.35	700	1.2309	0.6757	0.6806	0.6757	0.6446	0.7730
2.1878	3.59	750	1.1077	0.7169	0.7307	0.7169	0.6966	0.7999
2.1878	3.83	800	1.0393	0.7341	0.7548	0.7341	0.7226	0.8130
1.534	4.07	850	0.9478	0.7678	0.7794	0.7678	0.7572	0.8384
1.534	4.31	900	0.8755	0.7715	0.7789	0.7715	0.7627	0.8395
1.534	4.55	950	0.8563	0.7618	0.7737	0.7618	0.7491	0.8330
1.534	4.79	1000	0.7866	0.8007	0.8046	0.8007	0.7921	0.8616
1.2035	5.03	1050	0.7462	0.8007	0.8212	0.8007	0.7945	0.8591
1.2035	5.27	1100	0.7003	0.8157	0.8272	0.8157	0.8126	0.8717
1.2035	5.51	1150	0.7421	0.8105	0.8262	0.8105	0.8074	0.8672
1.2035	5.75	1200	0.7638	0.7993	0.8294	0.7993	0.7979	0.8595
1.2035	5.99	1250	0.6872	0.8187	0.8330	0.8187	0.8171	0.8742
0.9909	6.23	1300	0.6378	0.8345	0.8462	0.8345	0.8338	0.8840
0.9909	6.47	1350	0.6835	0.8075	0.8266	0.8075	0.8063	0.8669
0.9909	6.71	1400	0.6367	0.8345	0.8480	0.8345	0.8337	0.8874
0.9909	6.95	1450	0.5793	0.8434	0.8521	0.8434	0.8425	0.8931
0.8826	7.19	1500	0.6528	0.8307	0.8458	0.8307	0.8293	0.8824
0.8826	7.43	1550	0.6361	0.8225	0.8382	0.8225	0.8218	0.8761
0.8826	7.66	1600	0.6189	0.8360	0.8478	0.8360	0.8334	0.8855
0.8826	7.9	1650	0.6078	0.8337	0.8433	0.8337	0.8321	0.8831
0.7752	8.14	1700	0.6868	0.8315	0.8436	0.8315	0.8289	0.8835
0.7752	8.38	1750	0.6118	0.8419	0.8549	0.8419	0.8411	0.8897
0.7752	8.62	1800	0.5837	0.8532	0.8660	0.8532	0.8531	0.8974
0.7752	8.86	1850	0.5758	0.8487	0.8613	0.8487	0.8494	0.8956
0.7067	9.1	1900	0.6950	0.8307	0.8490	0.8307	0.8279	0.8827
0.7067	9.34	1950	0.5968	0.8479	0.8595	0.8479	0.8470	0.8942
0.7067	9.58	2000	0.5714	0.8614	0.8696	0.8614	0.8613	0.9035
0.7067	9.82	2050	0.6389	0.8427	0.8538	0.8427	0.8415	0.8903
0.6457	10.06	2100	0.6504	0.8502	0.8639	0.8502	0.8504	0.8948
0.6457	10.3	2150	0.5776	0.8547	0.8659	0.8547	0.8534	0.8988
0.6457	10.54	2200	0.6775	0.8434	0.8570	0.8434	0.8438	0.8912
0.6457	10.78	2250	0.5849	0.8569	0.8686	0.8569	0.8579	0.9013
0.6098	11.02	2300	0.5767	0.8622	0.8706	0.8622	0.8632	0.9037
0.6098	11.26	2350	0.6875	0.8404	0.8588	0.8404	0.8404	0.8898
0.6098	11.5	2400	0.7397	0.8352	0.8483	0.8352	0.8340	0.8865
0.6098	11.74	2450	0.5998	0.8629	0.8716	0.8629	0.8618	0.9053
0.6098	11.98	2500	0.6435	0.8449	0.8549	0.8449	0.8441	0.8918
0.5538	12.22	2550	0.6969	0.8502	0.8640	0.8502	0.8508	0.8965
0.5538	12.46	2600	0.6323	0.8577	0.8710	0.8577	0.8566	0.9006
0.5538	12.69	2650	0.6989	0.8532	0.8660	0.8532	0.8525	0.8981
0.5538	12.93	2700	0.6736	0.8554	0.8666	0.8554	0.8552	0.8994
0.5356	13.17	2750	0.6737	0.8487	0.8584	0.8487	0.8469	0.8960
0.5356	13.41	2800	0.6893	0.8457	0.8565	0.8457	0.8452	0.8921

Framework versions

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