aoi_clip_high_resolution_concate_fusin_crop_each_text

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: 3.4957
• Accuracy: 0.0648

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
1.7286	5.9821	1602	3.0151	0.0654
1.6207	11.9642	3204	3.2376	0.0665
1.5399	17.9462	4806	3.2386	0.0685
1.4981	23.9283	6408	3.3545	0.0673
1.4774	29.9104	8010	3.3404	0.0677
1.4648	35.8925	9612	3.4236	0.0670
1.4549	41.8745	11214	3.4689	0.0664
1.4528	47.8566	12816	3.5205	0.0659
1.4538	53.8387	14418	3.4703	0.0655
1.4519	59.8208	16020	3.4957	0.0651

Framework versions

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