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: 0.3785
• Wer Score: 2.7939

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: 10
• eval_batch_size: 10
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 40
• 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
6.818	4.4444	50	4.1955	4.1407
2.3051	8.8889	100	0.6527	2.5893
0.2976	13.3333	150	0.2576	2.2313
0.1103	17.7778	200	0.2680	2.9372
0.056	22.2222	250	0.3025	2.8351
0.0323	26.6667	300	0.3286	2.9704
0.0207	31.1111	350	0.3488	2.9758
0.0144	35.5556	400	0.3587	2.8506
0.0111	40.0	450	0.3688	2.7654
0.0091	44.4444	500	0.3762	2.8455
0.0081	48.8889	550	0.3785	2.7939

Framework versions

• Transformers 4.42.0.dev0
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1