lexical_classifier_bangla_assamese_v2

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on [https://huggingface.co/datasets/anrikus/lexical_diff_bangla_assamese_v2]. It achieves the following results on the evaluation set:

• Loss: 1.1317
• Accuracy: 0.7033
• Precision: 0.7480
• Recall: 0.6133
• F1: 0.6740

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: 1e-06
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.1501	1.0	35	1.0773	0.7286	0.7302	0.6866	0.7077
0.1789	2.0	70	0.8471	0.7714	0.7869	0.7164	0.7500
0.1463	3.0	105	1.3021	0.7071	0.7407	0.5970	0.6612
0.1664	4.0	140	1.1251	0.6857	0.6825	0.6418	0.6615
0.1746	5.0	175	0.8354	0.7214	0.7692	0.5970	0.6723
0.2282	6.0	210	1.2394	0.6929	0.6935	0.6418	0.6667
0.103	7.0	245	1.3935	0.6857	0.6769	0.6567	0.6667
0.186	8.0	280	1.1753	0.7	0.6812	0.7015	0.6912
0.2189	9.0	315	1.1819	0.6929	0.7069	0.6119	0.6560
0.1476	10.0	350	1.4722	0.65	0.65	0.5821	0.6142
0.2055	11.0	385	0.7656	0.7571	0.7895	0.6716	0.7258
0.1607	12.0	420	0.9756	0.7071	0.76	0.5672	0.6496
0.1532	13.0	455	0.9945	0.7071	0.7031	0.6716	0.6870
0.1023	14.0	490	1.1967	0.7071	0.7031	0.6716	0.6870
0.2389	15.0	525	0.7984	0.7643	0.7742	0.7164	0.7442
0.1925	16.0	560	0.9343	0.7143	0.7368	0.6269	0.6774
0.2038	17.0	595	1.1440	0.6857	0.6949	0.6119	0.6508
0.2193	18.0	630	0.9709	0.7071	0.7167	0.6418	0.6772
0.1719	19.0	665	0.9007	0.7429	0.7818	0.6418	0.7049
0.2334	20.0	700	0.8711	0.7429	0.7818	0.6418	0.7049
0.131	21.0	735	1.0785	0.7143	0.7288	0.6418	0.6825
0.2316	22.0	770	1.1080	0.6643	0.6786	0.5672	0.6179
0.1815	23.0	805	1.2657	0.6929	0.7308	0.5672	0.6387
0.1521	24.0	840	1.2584	0.7	0.6812	0.7015	0.6912
0.244	25.0	875	1.0375	0.7786	0.7812	0.7463	0.7634
0.3668	26.0	910	1.1253	0.7286	0.7458	0.6567	0.6984
0.1564	27.0	945	0.9891	0.7214	0.7414	0.6418	0.688
0.1782	28.0	980	0.9936	0.7357	0.75	0.6716	0.7087
0.1945	29.0	1015	0.9586	0.7357	0.7419	0.6866	0.7132
0.271	30.0	1050	0.8128	0.7357	0.7778	0.6269	0.6942
0.1889	31.0	1085	1.2141	0.6714	0.7059	0.5373	0.6102
0.1928	32.0	1120	1.0059	0.7143	0.7368	0.6269	0.6774
0.2035	33.0	1155	1.1185	0.6929	0.7069	0.6119	0.6560
0.226	34.0	1190	1.1719	0.6286	0.6271	0.5522	0.5873
0.1801	35.0	1225	1.1689	0.6786	0.6719	0.6418	0.6565
0.2353	36.0	1260	1.1392	0.7	0.6923	0.6716	0.6818
0.1686	37.0	1295	1.2064	0.6429	0.6667	0.5075	0.5763
0.2278	38.0	1330	0.8528	0.75	0.7759	0.6716	0.7200
0.1905	39.0	1365	1.2736	0.6643	0.6786	0.5672	0.6179
0.2136	40.0	1400	1.0255	0.7214	0.7333	0.6567	0.6929
0.1544	41.0	1435	0.9427	0.7214	0.7333	0.6567	0.6929
0.2691	42.0	1470	1.0433	0.7286	0.7544	0.6418	0.6935
0.2804	43.0	1505	1.2006	0.6929	0.7143	0.5970	0.6504
0.2345	44.0	1540	0.9487	0.75	0.7857	0.6567	0.7154
0.2541	45.0	1575	0.9468	0.7429	0.7246	0.7463	0.7353
0.2718	46.0	1610	1.3955	0.6714	0.6909	0.5672	0.6230
0.3179	47.0	1645	1.3356	0.6786	0.7037	0.5672	0.6281
0.4808	48.0	1680	0.9297	0.7429	0.7719	0.6567	0.7097
0.3231	49.0	1715	0.8732	0.7429	0.7818	0.6418	0.7049
0.3681	50.0	1750	1.0578	0.6857	0.7255	0.5522	0.6271

Framework versions

• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1