README.md

metadata

library_name: transformers
tags:
  - generated_from_trainer
model-index:
  - name: BengaliDiarization
    results: []

BengaliDiarization

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

• Loss: 0.2690
• Der: 0.0945
• False Alarm: 0.0140
• Missed Detection: 0.0450
• Confusion: 0.0355

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: 8e-05
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Der	False Alarm	Missed Detection	Confusion
0.5565	1.7241	100	0.3220	0.1152	0.0158	0.0628	0.0366
0.498	3.4483	200	0.3009	0.1055	0.0164	0.0474	0.0418
0.4955	5.1724	300	0.2863	0.0979	0.0161	0.0431	0.0387
0.4766	6.8966	400	0.2784	0.0981	0.0162	0.0429	0.0391
0.4726	8.6207	500	0.2755	0.0970	0.0151	0.0435	0.0385
0.4574	10.3448	600	0.2740	0.0966	0.0145	0.0450	0.0371
0.4441	12.0690	700	0.2719	0.0956	0.0143	0.0445	0.0369
0.4157	13.7931	800	0.2736	0.0958	0.0141	0.0449	0.0368
0.4916	15.5172	900	0.2722	0.0954	0.0141	0.0453	0.0361
0.4358	17.2414	1000	0.2682	0.0941	0.0141	0.0448	0.0352
0.4287	18.9655	1100	0.2690	0.0945	0.0140	0.0450	0.0355

Framework versions

• Transformers 4.57.1
• Pytorch 2.9.1+cu128
• Datasets 4.0.0
• Tokenizers 0.22.1