3adapter_backbone_100k_cls_content

This model is a fine-tuned version of ttqdunggg/3adapter_backbone_100k on the None dataset. It achieves the following results on the evaluation set:

• Acc Content: 0.9639
• F1 Content: 0.9366
• Acc Classification: 0.8401
• F1 Classification: 0.8136
• Loss: 0.2292

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
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Acc Content	F1 Content	Acc Classification	F1 Classification	Validation Loss
No log	1.0	276	0.9612	0.9319	0.8013	0.7655	0.2009
0.2424	2.0	552	0.9601	0.9306	0.8190	0.7896	0.1794
0.2424	3.0	828	0.9630	0.9350	0.8265	0.7943	0.1772
0.1156	4.0	1104	0.9623	0.9331	0.8381	0.8060	0.1805
0.1156	5.0	1380	0.9617	0.9329	0.8374	0.8106	0.1834
0.0695	6.0	1656	0.9619	0.9332	0.8419	0.8131	0.1952
0.0695	7.0	1932	0.9617	0.9312	0.8378	0.8102	0.2066
0.0401	8.0	2208	0.9617	0.9314	0.8383	0.8141	0.2164
0.0401	9.0	2484	0.9639	0.9368	0.8376	0.8120	0.2261
0.0246	10.0	2760	0.9639	0.9366	0.8401	0.8136	0.2292

Framework versions

• Transformers 4.57.1
• Pytorch 2.6.0+cu124
• Datasets 4.4.1
• Tokenizers 0.22.1