DD0101/disfluency_base_augmented_90_90

disfluency-large-3

This model is a fine-tuned version of vinai/phobert-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0403
• Precision: 0.9904
• Recall: 0.9880
• F1: 0.9892
• Accuracy: 0.9962

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: 5e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	280	0.0331	0.9719	0.9754	0.9736	0.9926
0.0853	2.0	560	0.0354	0.9771	0.9736	0.9753	0.9923
0.0853	3.0	840	0.0360	0.9759	0.9754	0.9757	0.9928
0.0119	4.0	1120	0.0255	0.9850	0.9838	0.9844	0.9948
0.0119	5.0	1400	0.0300	0.9873	0.9850	0.9862	0.9952
0.0063	6.0	1680	0.0412	0.9848	0.9742	0.9795	0.9927
0.0063	7.0	1960	0.0304	0.9844	0.9838	0.9841	0.9952
0.0039	8.0	2240	0.0344	0.9855	0.9820	0.9837	0.9939
0.004	9.0	2520	0.0522	0.9740	0.9681	0.9711	0.9911
0.004	10.0	2800	0.0305	0.9790	0.9790	0.9790	0.9943
0.0022	11.0	3080	0.0355	0.9837	0.9820	0.9829	0.9945
0.0022	12.0	3360	0.0400	0.9795	0.9772	0.9783	0.9935
0.002	13.0	3640	0.0394	0.9826	0.9814	0.9820	0.9943
0.002	14.0	3920	0.0452	0.9795	0.9772	0.9783	0.9930
0.0015	15.0	4200	0.0405	0.9825	0.9808	0.9817	0.9935
0.0015	16.0	4480	0.0373	0.9832	0.9826	0.9829	0.9941
0.0013	17.0	4760	0.0361	0.9832	0.9850	0.9841	0.9946
0.0013	18.0	5040	0.0447	0.9807	0.9790	0.9798	0.9937
0.0013	19.0	5320	0.0340	0.9874	0.9856	0.9865	0.9955
0.0009	20.0	5600	0.0374	0.9873	0.9826	0.9849	0.9948
0.0009	21.0	5880	0.0410	0.9843	0.9784	0.9813	0.9943
0.0007	22.0	6160	0.0275	0.9892	0.9862	0.9877	0.9961
0.0007	23.0	6440	0.0360	0.9891	0.9850	0.9871	0.9960
0.0011	24.0	6720	0.0323	0.9868	0.9850	0.9859	0.9954
0.0006	25.0	7000	0.0386	0.9867	0.9820	0.9843	0.9949
0.0006	26.0	7280	0.0408	0.9819	0.9802	0.9811	0.9940
0.0005	27.0	7560	0.0357	0.9867	0.9826	0.9846	0.9953
0.0005	28.0	7840	0.0370	0.9843	0.9820	0.9832	0.9946
0.0004	29.0	8120	0.0313	0.9880	0.9874	0.9877	0.9960
0.0004	30.0	8400	0.0363	0.9892	0.9862	0.9877	0.9956
0.0004	31.0	8680	0.0402	0.9843	0.9826	0.9835	0.9946
0.0004	32.0	8960	0.0321	0.9868	0.9850	0.9859	0.9956
0.0004	33.0	9240	0.0362	0.9861	0.9838	0.9850	0.9950
0.0003	34.0	9520	0.0307	0.9886	0.9880	0.9883	0.9964
0.0003	35.0	9800	0.0350	0.9880	0.9862	0.9871	0.9956
0.0001	36.0	10080	0.0343	0.9868	0.9856	0.9862	0.9956
0.0001	37.0	10360	0.0374	0.9874	0.9856	0.9865	0.9952
0.0003	38.0	10640	0.0333	0.9874	0.9868	0.9871	0.9957
0.0003	39.0	10920	0.0331	0.9886	0.9862	0.9874	0.9956
0.0001	40.0	11200	0.0349	0.9880	0.9868	0.9874	0.9961
0.0001	41.0	11480	0.0407	0.9880	0.9868	0.9874	0.9958
0.0001	42.0	11760	0.0389	0.9874	0.9868	0.9871	0.9959
0.0001	43.0	12040	0.0387	0.9892	0.9874	0.9883	0.9961
0.0001	44.0	12320	0.0414	0.9886	0.9868	0.9877	0.9959
0.0001	45.0	12600	0.0386	0.9886	0.9868	0.9877	0.9961
0.0001	46.0	12880	0.0408	0.9892	0.9874	0.9883	0.9961
0.0	47.0	13160	0.0402	0.9898	0.9880	0.9889	0.9962
0.0	48.0	13440	0.0411	0.9886	0.9868	0.9877	0.9959
0.0	49.0	13720	0.0403	0.9904	0.9880	0.9892	0.9962
0.0	50.0	14000	0.0402	0.9904	0.9880	0.9892	0.9962

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.1
• Tokenizers 0.13.3