validated_cracks_dataset_relabeled_convnext_model

This model is a fine-tuned version of facebook/convnext-base-224-22k on the Logiroad/validated_cracks_dataset_relabeled dataset. It achieves the following results on the evaluation set:

• Loss: 0.9692
• Accuracy: 0.6332

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 1337
• 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: 0.05
• num_epochs: 30.0
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.05

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.3724	1.0	108	1.3528	0.3359
1.3528	2.0	216	1.2849	0.4208
1.2839	3.0	324	1.2019	0.4633
1.2608	4.0	432	1.1560	0.5212
1.2178	5.0	540	1.0907	0.5560
1.1624	6.0	648	1.0543	0.5290
1.1108	7.0	756	1.0452	0.5598
1.1028	8.0	864	1.0569	0.5598
1.1023	9.0	972	1.0580	0.5676
1.0572	10.0	1080	1.0031	0.6139
0.9874	11.0	1188	1.0340	0.5830
1.0132	12.0	1296	1.0049	0.6100
0.9898	13.0	1404	0.9875	0.6216
1.0182	14.0	1512	0.9668	0.6100
0.9889	15.0	1620	0.9692	0.6332
0.9446	16.0	1728	0.9777	0.6332
0.9519	17.0	1836	1.0030	0.5985
0.9458	18.0	1944	0.9748	0.5985
0.9347	19.0	2052	0.9744	0.6178
0.8863	20.0	2160	0.9657	0.6293
0.8507	21.0	2268	0.9784	0.6255
0.8712	22.0	2376	0.9790	0.6255
0.8857	23.0	2484	0.9682	0.6178
0.8848	24.0	2592	0.9723	0.6255
0.904	25.0	2700	0.9754	0.6178
0.943	26.0	2808	0.9710	0.6216
0.862	27.0	2916	0.9717	0.6178
0.864	28.0	3024	0.9705	0.6178
0.8879	29.0	3132	0.9703	0.6178
0.8099	30.0	3240	0.9703	0.6178

Framework versions

• Transformers 4.46.1
• Pytorch 2.3.0
• Datasets 3.1.0
• Tokenizers 0.20.3