git-base-food

This model is a fine-tuned version of microsoft/git-base on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.0444
• Wer Score: 10.0470

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Score
No log	1.05	20	7.3069	113.3758
No log	2.11	40	5.3566	3.2282
No log	3.16	60	3.4409	1.1879
No log	4.21	80	1.7218	1.1007
No log	5.26	100	0.5834	1.0872
No log	6.32	120	0.1684	1.3020
No log	7.37	140	0.0720	2.9732
No log	8.42	160	0.0507	2.0805
No log	9.47	180	0.0467	3.0336
No log	10.53	200	0.0415	10.6107
No log	11.58	220	0.0425	7.7383
No log	12.63	240	0.0426	14.1745
No log	13.68	260	0.0434	6.0067
No log	14.74	280	0.0447	10.5503
No log	15.79	300	0.0434	9.1678
No log	16.84	320	0.0439	10.8591
No log	17.89	340	0.0446	10.0470
No log	18.95	360	0.0444	10.1208
No log	20.0	380	0.0444	10.0470

Framework versions

• Transformers 4.34.1
• Pytorch 2.1.0+cu118
• Datasets 2.14.5
• Tokenizers 0.14.1