bartpho-word-base-ed-multi-v3

This model is a fine-tuned version of vinai/bartpho-word-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0236
• F1 Micro: 0.7990
• Recall Micro: 0.7985
• Precision Micro: 0.7996
• F1 Macro: 0.6436
• Recall Macro: 0.6553
• Precision Macro: 0.6605

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Micro	Recall Micro	Precision Micro	F1 Macro	Recall Macro	Precision Macro
No log	0.9987	393	0.0267	0.7802	0.7647	0.7963	0.5052	0.4675	0.6180
0.1473	2.0	787	0.0268	0.7717	0.7827	0.7610	0.6187	0.6788	0.5931
0.0245	2.9987	1180	0.0232	0.7970	0.7898	0.8044	0.5962	0.5898	0.6782
0.0193	4.0	1574	0.0224	0.8015	0.7964	0.8066	0.6273	0.6125	0.6827
0.0193	4.9936	1965	0.0236	0.7990	0.7985	0.7996	0.6436	0.6553	0.6605

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1