This model is a fine-tuned version of google-bert/bert-base-multilingual-uncased on the Custom Devangari Datasets dataset. It achieves the following results on the evaluation set:

Loss: 0.6887
Accuracy: 0.8660
F1: 0.4658
Precision: 0.4343
Recall: 0.5021

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: 32
eval_batch_size: 32
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 5
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.5999	1.0	595	0.5313	0.7274	0.3965	0.2670	0.7700
0.5114	2.0	1190	0.4717	0.7745	0.4427	0.3106	0.7700
0.4005	3.0	1785	0.4986	0.7907	0.4556	0.3266	0.7532
0.3087	4.0	2380	0.6887	0.8660	0.4658	0.4343	0.5021
0.2292	5.0	2975	0.8148	0.8626	0.4615	0.4240	0.5063

Framework versions

Transformers 4.44.2
Pytorch 2.4.1+cu121
Datasets 3.0.2
Tokenizers 0.19.1

luluw/Nepali-BERT-sentiment

Nepali-BERT-sentiment

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

luluw/
Nepali-BERT-sentiment