output_diff_approach

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

• Loss: 0.1603
• 5 Err Precision: 0.0
• 5 Err Recall: 0.0
• 5 Err F1: 0.0
• 5 Err Number: 34
• Precision: 0.4328
• Recall: 0.3244
• F1: 0.3709
• Number: 9934
• Err Precision: 0.0
• Err Recall: 0.0
• Err F1: 0.0
• Err Number: 285
• Egin Err Precision: 0.7528
• Egin Err Recall: 0.4192
• Egin Err F1: 0.5385
• Egin Err Number: 1126
• El Err Precision: 0.7112
• El Err Recall: 0.2891
• El Err F1: 0.4111
• El Err Number: 1380
• Nd Err Precision: 0.6986
• Nd Err Recall: 0.4487
• Nd Err F1: 0.5464
• Nd Err Number: 1188
• Ne Word Err Precision: 0.7223
• Ne Word Err Recall: 0.6297
• Ne Word Err F1: 0.6728
• Ne Word Err Number: 8247
• Unc Insert Err Precision: 0.6140
• Unc Insert Err Recall: 0.0776
• Unc Insert Err F1: 0.1378
• Unc Insert Err Number: 902
• Micro Avg Precision: 0.5922
• Micro Avg Recall: 0.4282
• Micro Avg F1: 0.4970
• Micro Avg Number: 23096
• Macro Avg Precision: 0.4915
• Macro Avg Recall: 0.2736
• Macro Avg F1: 0.3347
• Macro Avg Number: 23096
• Weighted Avg Precision: 0.5832
• Weighted Avg Recall: 0.4282
• Weighted Avg F1: 0.4841
• Weighted Avg Number: 23096
• Overall Accuracy: 0.9514

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
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 2.0

Training results

Training Loss	Epoch	Step	Validation Loss	5 Err Precision	5 Err Recall	5 Err F1	5 Err Number	Precision	Recall	F1	Number	Err Precision	Err Recall	Err F1	Err Number	Egin Err Precision	Egin Err Recall	Egin Err F1	Egin Err Number	El Err Precision	El Err Recall	El Err F1	El Err Number	Nd Err Precision	Nd Err Recall	Nd Err F1	Nd Err Number	Ne Word Err Precision	Ne Word Err Recall	Ne Word Err F1	Ne Word Err Number	Unc Insert Err Precision	Unc Insert Err Recall	Unc Insert Err F1	Unc Insert Err Number	Micro Avg Precision	Micro Avg Recall	Micro Avg F1	Micro Avg Number	Macro Avg Precision	Macro Avg Recall	Macro Avg F1	Macro Avg Number	Weighted Avg Precision	Weighted Avg Recall	Weighted Avg F1	Weighted Avg Number	Overall Accuracy
0.3987	1.0	575	0.1930	0.0	0.0	0.0	34	0.3517	0.1737	0.2326	9934	0.0	0.0	0.0	285	0.8127	0.2389	0.3693	1126	0.8345	0.1681	0.2799	1380	0.6727	0.3460	0.4569	1188	0.7470	0.4215	0.5389	8247	0.25	0.0011	0.0022	902	0.5670	0.2648	0.3610	23096	0.4586	0.1687	0.2350	23096	0.5519	0.2648	0.3508	23096	0.9422
0.1861	2.0	1150	0.1603	0.0	0.0	0.0	34	0.4328	0.3244	0.3709	9934	0.0	0.0	0.0	285	0.7528	0.4192	0.5385	1126	0.7112	0.2891	0.4111	1380	0.6986	0.4487	0.5464	1188	0.7223	0.6297	0.6728	8247	0.6140	0.0776	0.1378	902	0.5922	0.4282	0.4970	23096	0.4915	0.2736	0.3347	23096	0.5832	0.4282	0.4841	23096	0.9514

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.1+cu117
• Datasets 2.9.0
• Tokenizers 0.13.2