fotocopy-ori

This model is a fine-tuned version of openai/clip-vit-base-patch32 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.4776
• Accuracy: 0.9492

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.9231	3	0.6343	0.4915
No log	1.8462	6	0.2235	0.9322
No log	2.7692	9	0.1887	0.9492
No log	4.0	13	0.0278	0.9831
0.4375	4.9231	16	1.6119	0.8475
0.4375	5.8462	19	0.5158	0.8983
0.4375	6.7692	22	0.0602	0.9661
0.4375	8.0	26	0.3831	0.9492
0.4375	8.9231	29	0.4555	0.9492
0.1245	9.8462	32	0.9890	0.9153
0.1245	10.7692	35	0.4632	0.9322
0.1245	12.0	39	0.5992	0.9322
0.1245	12.9231	42	0.6255	0.9322
0.048	13.8462	45	0.5156	0.9492
0.048	14.7692	48	0.6033	0.9492
0.048	16.0	52	0.5978	0.9492
0.048	16.9231	55	0.5747	0.9492
0.048	17.8462	58	0.5635	0.9492
0.0005	18.7692	61	0.5314	0.9492
0.0005	20.0	65	0.5023	0.9492
0.0005	20.9231	68	0.4886	0.9492
0.0005	21.8462	71	0.4809	0.9492
0.0005	22.7692	74	0.4779	0.9492
0.0	23.0769	75	0.4776	0.9492

Framework versions

• Transformers 4.41.2
• Pytorch 2.1.2
• Datasets 2.19.2
• Tokenizers 0.19.1