aoi_clip_high_resolution_concate_fusin_crop_each_text_512

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: 3.5539
• Accuracy: 0.0669

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: 15
• eval_batch_size: 20
• seed: 42
• gradient_accumulation_steps: 14
• total_train_batch_size: 210
• 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
1.5814	6.0	1530	3.0257	0.0726
1.4807	12.0	3060	3.2677	0.0712
1.4075	18.0	4590	3.3332	0.0703
1.3618	24.0	6120	3.2491	0.0692
1.3396	30.0	7650	3.3756	0.0690
1.3298	36.0	9180	3.5386	0.0678
1.324	42.0	10710	3.5245	0.0675
1.3177	48.0	12240	3.5136	0.0671
1.3181	54.0	13770	3.4984	0.0669
1.3117	60.0	15300	3.5539	0.0669

Framework versions

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