roberta-base-downstream-indian-ner

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

• Loss: 0.2666
• Precision: 0.5248
• Recall: 0.7557
• F1: 0.6195
• Accuracy: 0.9547

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: 3e-05
• train_batch_size: 128
• eval_batch_size: 128
• seed: 1
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	86	0.3551	0.0892	0.4171	0.1469	0.7997
No log	2.0	172	0.2383	0.1328	0.4684	0.2070	0.8327
No log	3.0	258	0.2159	0.2075	0.5253	0.2975	0.8922
No log	4.0	344	0.2013	0.2338	0.5344	0.3253	0.9025
No log	5.0	430	0.1926	0.2732	0.5476	0.3646	0.9131
0.396	6.0	516	0.2002	0.2821	0.5717	0.3778	0.9134
0.396	7.0	602	0.2103	0.3407	0.6220	0.4403	0.9267
0.396	8.0	688	0.1944	0.3388	0.6265	0.4398	0.9256
0.396	9.0	774	0.2118	0.3477	0.6349	0.4494	0.9291
0.396	10.0	860	0.2274	0.4096	0.6729	0.5092	0.9396
0.396	11.0	946	0.2318	0.4527	0.7047	0.5513	0.9450
0.0715	12.0	1032	0.2439	0.4436	0.6946	0.5414	0.9443
0.0715	13.0	1118	0.2385	0.4781	0.7379	0.5802	0.9460
0.0715	14.0	1204	0.2420	0.4584	0.7065	0.5560	0.9460
0.0715	15.0	1290	0.2455	0.4992	0.7344	0.5944	0.9502
0.0715	16.0	1376	0.2513	0.5377	0.7644	0.6313	0.9572
0.0715	17.0	1462	0.2670	0.5354	0.7627	0.6291	0.9558
0.0344	18.0	1548	0.2687	0.5020	0.7351	0.5966	0.9505
0.0344	19.0	1634	0.2666	0.5248	0.7557	0.6195	0.9547

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1