still-universe-209

This model is a fine-tuned version of facebook/convnextv2-tiny-1k-224 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5721
• Accuracy: 0.6497
• Precision: 0.6965
• Recall: 0.6497
• F1: 0.6583
• Roc Auc: 0.8795

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: 0.0001
• train_batch_size: 256
• eval_batch_size: 256
• 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_ratio: 0.1
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Roc Auc
1.379	1.0	17	1.3021	0.4544	0.5543	0.4544	0.4617	0.7377
1.3695	2.0	34	1.2286	0.5391	0.5315	0.5391	0.5266	0.7806
1.1945	3.0	51	1.1127	0.5794	0.5651	0.5794	0.5647	0.8098
1.0517	4.0	68	0.8318	0.5872	0.6134	0.5872	0.5961	0.8273
1.0323	5.0	85	0.8958	0.5156	0.6297	0.5156	0.5319	0.8189
0.9029	6.0	102	0.7313	0.5365	0.6126	0.5365	0.5398	0.8267
0.9002	7.0	119	0.7217	0.5794	0.5998	0.5794	0.5558	0.8445
0.7855	8.0	136	0.6522	0.6029	0.6629	0.6029	0.6064	0.8581
0.756	9.0	153	0.6371	0.5964	0.6263	0.5964	0.5653	0.8643
0.7164	10.0	170	0.6291	0.5690	0.6930	0.5690	0.5780	0.8579
0.6894	11.0	187	0.6194	0.5938	0.6360	0.5938	0.5735	0.8699
0.6606	12.0	204	0.5834	0.6289	0.6906	0.6289	0.6402	0.8742
0.6273	13.0	221	0.5766	0.6510	0.6972	0.6510	0.6607	0.8780
0.6046	14.0	238	0.5732	0.6497	0.6965	0.6497	0.6583	0.8790
0.6255	15.0	255	0.5721	0.6497	0.6965	0.6497	0.6583	0.8795

Framework versions

• Transformers 4.52.3
• Pytorch 2.7.0+cpu
• Datasets 3.6.0
• Tokenizers 0.21.0