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: 0.1817
• Wer Score: 9.0938

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

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Score
7.3974	0.7	50	4.5248	4.5234
2.2794	1.4	100	0.4021	5.1680
0.1697	2.1	150	0.1398	1.5039
0.0816	2.8	200	0.1458	9.9570
0.0556	3.5	250	0.1417	2.5234
0.043	4.2	300	0.1448	12.8086
0.0285	4.9	350	0.1469	7.3867
0.021	5.59	400	0.1505	13.0312
0.0205	6.29	450	0.1499	6.3281
0.0179	6.99	500	0.1527	13.0234
0.0157	7.69	550	0.1552	6.3047
0.015	8.39	600	0.1571	6.7656
0.015	9.09	650	0.1579	10.2305
0.0137	9.79	700	0.1585	11.4219
0.0132	10.49	750	0.1598	5.8320
0.0132	11.19	800	0.1591	12.0508
0.013	11.89	850	0.1612	7.9492
0.0117	12.59	900	0.1621	8.1758
0.0123	13.29	950	0.1632	12.9961
0.0125	13.99	1000	0.1613	10.2031
0.0116	14.69	1050	0.1642	5.7930
0.0112	15.38	1100	0.1636	6.1719
0.0112	16.08	1150	0.1652	7.2422
0.0107	16.78	1200	0.1644	12.9961
0.0108	17.48	1250	0.1661	5.0117
0.0109	18.18	1300	0.1658	7.3242
0.0108	18.88	1350	0.1691	6.0547
0.0101	19.58	1400	0.1690	6.9141
0.0103	20.28	1450	0.1692	7.1680
0.0107	20.98	1500	0.1702	12.3281
0.0099	21.68	1550	0.1708	10.75
0.0103	22.38	1600	0.1714	9.5586
0.0101	23.08	1650	0.1713	12.9805
0.0098	23.78	1700	0.1712	11.4883
0.0095	24.48	1750	0.1711	9.3320
0.0096	25.17	1800	0.1738	8.6523
0.0097	25.87	1850	0.1717	11.5078
0.0091	26.57	1900	0.1735	7.9570
0.0092	27.27	1950	0.1729	9.8242
0.0093	27.97	2000	0.1721	10.5078
0.0087	28.67	2050	0.1732	9.3906
0.009	29.37	2100	0.1760	8.0664
0.009	30.07	2150	0.1769	10.5312
0.0086	30.77	2200	0.1743	10.8555
0.0087	31.47	2250	0.1772	10.2188
0.0089	32.17	2300	0.1757	11.6016
0.0088	32.87	2350	0.1765	8.9297
0.0082	33.57	2400	0.1754	9.6484
0.0082	34.27	2450	0.1770	12.3711
0.0084	34.97	2500	0.1761	10.1523
0.0076	35.66	2550	0.1774	9.1055
0.0077	36.36	2600	0.1788	8.7852
0.0079	37.06	2650	0.1782	11.8086
0.0071	37.76	2700	0.1784	10.5234
0.0075	38.46	2750	0.1789	8.8828
0.0072	39.16	2800	0.1796	8.5664
0.0071	39.86	2850	0.1804	9.5391
0.0069	40.56	2900	0.1796	9.4062
0.0068	41.26	2950	0.1797	8.9883
0.0067	41.96	3000	0.1809	10.5273
0.0062	42.66	3050	0.1801	10.4531
0.0062	43.36	3100	0.1803	7.2188
0.0063	44.06	3150	0.1808	8.7930
0.0058	44.76	3200	0.1804	10.5156
0.0057	45.45	3250	0.1807	11.1328
0.0059	46.15	3300	0.1812	8.6875
0.0055	46.85	3350	0.1811	10.2773
0.0053	47.55	3400	0.1814	10.0391
0.0054	48.25	3450	0.1817	8.5391
0.0053	48.95	3500	0.1818	8.9688
0.005	49.65	3550	0.1817	9.0938

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3