Modelpublic

levshechter/tibetan_code_switching_model

sourceHugging Faceupdated 1y agoView on Hugging Face

0likes10downloads

Model Card

tibetancodeswitching_model

This model is a fine-tuned version of OMRIDRORI/mbert-tibetan-continual-unicode-240k on an unknown dataset. It achieves the following results on the evaluation set:

—Loss: 0.3714
—Accuracy: 0.9445
—Switch Precision: 0.9381
—Switch Recall: 0.9725
—Switch F1: 0.9550
—True Switches: 109
—Pred Switches: 113
—Tp: 106
—Fp: 7
—Fn: 3

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
—trainbatchsize: 8
—evalbatchsize: 8
—seed: 42
—optimizer: Use adamwtorch with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
—lrschedulertype: linear
—lrschedulerwarmup_steps: 100
—num_epochs: 15
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Switch Precision	Switch Recall	Switch F1	True Switches	Pred Switches	Tp	Fp	Fn
4.727	1.0	50	5.6020	0.2133	0.0749	0.9908	0.1393	109	1442	108	1334	1
1.7867	2.0	100	1.2875	0.5304	0.5365	0.9450	0.6844	109	192	103	89	6
1.1232	3.0	150	0.9034	0.6356	0.6835	0.9908	0.8090	109	158	108	50	1
0.7733	4.0	200	0.6946	0.7664	0.6855	1.0	0.8134	109	159	109	50	0
0.5582	5.0	250	0.4907	0.8514	0.7219	1.0	0.8385	109	151	109	42	0
0.4037	6.0	300	0.4428	0.8808	0.7219	1.0	0.8385	109	151	109	42	0
0.2593	7.0	350	0.3282	0.9213	0.8134	1.0	0.8971	109	134	109	25	0
0.1714	8.0	400	0.3918	0.9044	0.8244	0.9908	0.9000	109	131	108	23	1
0.1114	9.0	450	0.3473	0.9334	0.8710	0.9908	0.9270	109	124	108	16	1
0.0678	10.0	500	0.3275	0.9348	0.9145	0.9817	0.9469	109	117	107	10	2
0.075	11.0	550	0.3718	0.9392	0.8917	0.9817	0.9345	109	120	107	13	2
0.0276	12.0	600	0.3588	0.9435	0.9068	0.9817	0.9427	109	118	107	11	2
0.0251	13.0	650	0.3488	0.9411	0.9304	0.9817	0.9554	109	115	107	8	2
0.0154	14.0	700	0.3599	0.9455	0.9386	0.9817	0.9596	109	114	107	7	2
0.0165	15.0	750	0.3714	0.9445	0.9381	0.9725	0.9550	109	113	106	7	3

Framework versions

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