DipsankarSinha/bangla-sentiment-analysis-v2_2025

repo_name

This model is a fine-tuned version of csebuetnlp/banglabert on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4297
• Accuracy: 0.8705

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 645
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.3313	0.4735	1000	0.3475	0.8446
0.3384	0.9470	2000	0.3331	0.8630
0.2469	1.4205	3000	0.3431	0.8615
0.2392	1.8939	4000	0.3347	0.8705
0.1737	2.3674	5000	0.4186	0.8659
0.1481	2.8409	6000	0.4297	0.8705

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.4.1
• Tokenizers 0.21.1