tibetan-CS-detector_mbert-tibetan-continual-wylie_MUL_SEG_RUNI

This model is a fine-tuned version of OMRIDRORI/mbert-tibetan-continual-wylie-final on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: inf
• Accuracy: 0.9188
• Switch F1: 0.7457
• Switch Precision: 0.6681
• Switch Recall: 0.8436
• True Switches: 179
• Pred Switches: 226

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Switch F1	Switch Precision	Switch Recall	True Switches	Pred Switches
1.3462	0.5263	30	0.8696	0.8328	0.0	0.0	0.0	179	0
0.3733	1.0526	60	inf	0.8835	0.4703	0.4555	0.4860	179	191
0.0	1.5789	90	inf	0.8978	0.5523	0.3985	0.8994	179	404
0.1859	2.1053	120	inf	0.9083	0.5699	0.4220	0.8771	179	372
0.1912	2.6316	150	inf	0.9166	0.5029	0.3406	0.9609	179	505
0.1074	3.1579	180	inf	0.9056	0.5882	0.4319	0.9218	179	382
0.1625	3.6842	210	inf	0.9135	0.6025	0.4462	0.9274	179	372
0.2196	4.2105	240	inf	0.9174	0.6507	0.5062	0.9106	179	322
0.0967	4.7368	270	inf	0.9142	0.6595	0.5347	0.8603	179	288
0.0862	5.2632	300	inf	0.9114	0.7173	0.6749	0.7654	179	203
0.0249	5.7895	330	inf	0.9182	0.7002	0.5930	0.8547	179	258
0.099	6.3158	360	inf	0.9136	0.7223	0.6061	0.8939	179	264
0.046	6.8421	390	inf	0.9201	0.7291	0.6520	0.8268	179	227
0.0136	7.3684	420	inf	0.9183	0.7100	0.6071	0.8547	179	252
0.0973	7.8947	450	inf	0.9204	0.7482	0.6652	0.8547	179	230
0.038	8.4211	480	inf	0.9199	0.7464	0.6527	0.8715	179	239
0.0672	8.9474	510	inf	0.9208	0.7506	0.6726	0.8492	179	226
0.0139	9.4737	540	inf	0.9171	0.7363	0.6405	0.8659	179	242
0.018	10.0	570	inf	0.9188	0.7457	0.6681	0.8436	179	226

Framework versions

• Transformers 4.46.3
• Pytorch 2.4.1+cu121
• Datasets 2.0.0
• Tokenizers 0.20.3