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: 0.1481
• Wer Score: 7.2150

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• 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
0.0359	3.12	50	0.1192	0.8131
0.0174	6.25	100	0.1257	3.0654
0.0132	9.38	150	0.1283	0.7850
0.011	12.5	200	0.1297	1.4112
0.0095	15.62	250	0.1332	5.1028
0.0083	18.75	300	0.1376	5.5701
0.0077	21.88	350	0.1368	0.7944
0.0068	25.0	400	0.1366	5.6168
0.0061	28.12	450	0.1417	4.4299
0.0057	31.25	500	0.1406	6.6636
0.0047	34.38	550	0.1438	7.3738
0.0038	37.5	600	0.1448	7.6262
0.0032	40.62	650	0.1468	9.0841
0.0027	43.75	700	0.1473	6.8598
0.0024	46.88	750	0.1480	7.3178
0.0021	50.0	800	0.1481	7.2150

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1