git-base-pokemon

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: 0.0350
• Wer Score: 2.2148

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Score
7.3616	4.17	50	4.5895	21.4258
2.4353	8.33	100	0.4961	9.9322
0.1527	12.5	150	0.0303	1.3197
0.0192	16.67	200	0.0260	1.3299
0.007	20.83	250	0.0297	2.2059
0.0027	25.0	300	0.0321	2.4795
0.0017	29.17	350	0.0334	2.4488
0.0014	33.33	400	0.0340	2.1355
0.0013	37.5	450	0.0345	2.3619
0.0012	41.67	500	0.0349	2.2084
0.0011	45.83	550	0.0350	2.1803
0.0011	50.0	600	0.0350	2.2148

Framework versions

• Transformers 4.26.0
• Pytorch 1.13.1+cu116
• Datasets 2.9.0
• Tokenizers 0.13.2