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: 0.0396
• Wer Score: 6.0488

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: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Score
7.3484	1.06	50	4.4320	10.6547
2.1536	2.13	100	0.2910	1.8947
0.0909	3.19	150	0.0322	0.3684
0.0278	4.26	200	0.0275	0.3659
0.0211	5.32	250	0.0271	0.8858
0.0185	6.38	300	0.0267	0.6778
0.0155	7.45	350	0.0272	7.8190
0.0129	8.51	400	0.0279	3.2452
0.0108	9.57	450	0.0280	15.0462
0.0082	10.64	500	0.0291	10.0372
0.0069	11.7	550	0.0303	15.1592
0.0048	12.77	600	0.0321	15.4493
0.0033	13.83	650	0.0322	16.2439
0.0022	14.89	700	0.0350	17.7125
0.0017	15.96	750	0.0340	16.8357
0.0011	17.02	800	0.0354	16.8780
0.0009	18.09	850	0.0351	17.3273
0.0006	19.15	900	0.0364	16.4788
0.0005	20.21	950	0.0368	15.4442
0.0004	21.28	1000	0.0368	16.2336
0.0004	22.34	1050	0.0375	14.1168
0.0004	23.4	1100	0.0375	14.4365
0.0004	24.47	1150	0.0373	12.3890
0.0004	25.53	1200	0.0379	8.7843
0.0004	26.6	1250	0.0382	9.2298
0.0003	27.66	1300	0.0383	8.8562
0.0003	28.72	1350	0.0384	9.5777
0.0003	29.79	1400	0.0383	8.6021
0.0003	30.85	1450	0.0387	7.9782
0.0003	31.91	1500	0.0387	7.7394
0.0003	32.98	1550	0.0388	7.6431
0.0003	34.04	1600	0.0389	6.9037
0.0003	35.11	1650	0.0391	6.8665
0.0003	36.17	1700	0.0392	6.0526
0.0003	37.23	1750	0.0394	5.6996
0.0003	38.3	1800	0.0393	6.1361
0.0003	39.36	1850	0.0394	5.9127
0.0003	40.43	1900	0.0394	5.6816
0.0003	41.49	1950	0.0394	5.3723
0.0003	42.55	2000	0.0395	4.8806
0.0002	43.62	2050	0.0395	6.9178
0.0002	44.68	2100	0.0395	6.2953
0.0002	45.74	2150	0.0395	6.1142
0.0002	46.81	2200	0.0396	6.0642
0.0002	47.87	2250	0.0396	6.0077
0.0002	48.94	2300	0.0396	6.0026
0.0002	50.0	2350	0.0396	6.0488

Framework versions

• Transformers 4.29.2
• Pytorch 2.0.1+cu117
• Datasets 2.12.0
• Tokenizers 0.13.3