quickdraw-ViT-base-finetune

This model is a fine-tuned version of WinKawaks/vit-tiny-patch16-224 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8260
• Accuracy: 0.7892

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: 0.0008
• train_batch_size: 512
• eval_batch_size: 512
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10000
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.3104	0.5688	5000	1.2637	0.6826
1.1479	1.1377	10000	1.1421	0.7096
1.0236	1.7065	15000	1.0128	0.7404
0.9206	2.2753	20000	0.9457	0.7577
0.8878	2.8441	25000	0.9111	0.7652
0.8107	3.4130	30000	0.8754	0.7749
0.7874	3.9818	35000	0.8436	0.7827
0.7064	4.5506	40000	0.8360	0.7869

Framework versions

• Transformers 4.40.2
• Pytorch 2.2.1
• Datasets 2.19.1
• Tokenizers 0.19.1