Modelpublic

MHGanainy/roberta-base-legal-indian-courts-downstream-ildc

sourceHugging Faceupdated 2y agoView on Hugging Face

0likes22downloads

Model Card

roberta-base-legal-indian-courts-downstream-ildc

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

—Loss: 0.6117
—Accuracy: 0.8048
—Precision: 0.7875
—Recall: 0.8350
—F1: 0.8105
—Best Threshold: 0.1230

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
—trainbatchsize: 16
—evalbatchsize: 16
—seed: 1
—distributed_type: multi-GPU
—num_devices: 8
—totaltrainbatch_size: 128
—totalevalbatch_size: 128
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 20.0
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Best Threshold
No log	1.0	253	0.6666	0.5433	0.5237	0.9577	0.6771	0.3772
0.6458	2.0	506	0.4511	0.8008	0.8045	0.7948	0.7996	0.4237
0.6458	3.0	759	0.4500	0.8169	0.8082	0.8310	0.8194	0.3248
0.4793	4.0	1012	0.3898	0.8400	0.8153	0.8793	0.8461	0.3844
0.4793	5.0	1265	0.4364	0.8068	0.7537	0.9115	0.8251	0.1992
0.3786	6.0	1518	0.4749	0.8068	0.7572	0.9034	0.8239	0.1452
0.3786	7.0	1771	0.6117	0.8048	0.7875	0.8350	0.8105	0.1230

Framework versions

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