aoi_clip_high_resolution_concate_fusin_crop_each_text_256

This model is a fine-tuned version of OFA-Sys/chinese-clip-vit-base-patch16 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.5401
• Accuracy: 0.0635

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: 20
• eval_batch_size: 20
• seed: 42
• gradient_accumulation_steps: 10
• total_train_batch_size: 200
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 60.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.7271	5.9821	1602	3.0070	0.0789
1.6271	11.9642	3204	3.1514	0.0732
1.552	17.9462	4806	3.1511	0.0705
1.5094	23.9283	6408	3.3706	0.0684
1.484	29.9104	8010	3.4197	0.0672
1.4683	35.8925	9612	3.5270	0.0666
1.4567	41.8745	11214	3.4933	0.0658
1.4541	47.8566	12816	3.4874	0.0653
1.4539	53.8387	14418	3.5305	0.0646
1.452	59.8208	16020	3.5401	0.0640

Framework versions

• Transformers 4.42.3
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1