git-base-coco-pokemon

This model is a fine-tuned version of microsoft/git-base-coco on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0496
• Wer Score: 17.2413

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
• 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
No log	0.53	50	4.1410	21.7574
No log	1.06	100	0.2080	0.4284
No log	1.6	150	0.0389	0.4503
No log	2.13	200	0.0314	0.4297
No log	2.66	250	0.0302	0.3935
No log	3.19	300	0.0297	3.6761
No log	3.72	350	0.0302	0.9729
No log	4.26	400	0.0294	3.9110
No log	4.79	450	0.0296	0.6968
0.9165	5.32	500	0.0307	14.7252
0.9165	5.85	550	0.0306	14.0787
0.9165	6.38	600	0.0314	17.9974
0.9165	6.91	650	0.0305	19.4271
0.9165	7.45	700	0.0313	18.76
0.9165	7.98	750	0.0321	17.1084
0.9165	8.51	800	0.0322	20.9123
0.9165	9.04	850	0.0321	21.0026
0.9165	9.57	900	0.0328	19.2103
0.9165	10.11	950	0.0336	19.4503
0.0124	10.64	1000	0.0354	19.6310
0.0124	11.17	1050	0.0350	17.3652
0.0124	11.7	1100	0.0355	18.2955
0.0124	12.23	1150	0.0375	19.4194
0.0124	12.77	1200	0.0362	18.2606
0.0124	13.3	1250	0.0375	19.8348
0.0124	13.83	1300	0.0380	18.6581
0.0124	14.36	1350	0.0383	19.2723
0.0124	14.89	1400	0.0407	18.8516
0.0124	15.43	1450	0.0406	19.0968
0.0049	15.96	1500	0.0406	18.4774
0.0049	16.49	1550	0.0419	18.5768
0.0049	17.02	1600	0.0435	19.8606
0.0049	17.55	1650	0.0437	19.6477
0.0049	18.09	1700	0.0445	19.2684
0.0049	18.62	1750	0.0443	18.6039
0.0049	19.15	1800	0.0432	17.8129
0.0049	19.68	1850	0.0455	18.9587
0.0049	20.21	1900	0.0448	18.28
0.0049	20.74	1950	0.0455	18.4477
0.0009	21.28	2000	0.0453	18.2542
0.0009	21.81	2050	0.0457	18.7458
0.0009	22.34	2100	0.0456	18.5239
0.0009	22.87	2150	0.0450	18.3523
0.0009	23.4	2200	0.0459	18.2658
0.0009	23.94	2250	0.0462	18.0916
0.0009	24.47	2300	0.0465	18.3265
0.0009	25.0	2350	0.0463	18.4245
0.0009	25.53	2400	0.0466	18.1948
0.0009	26.06	2450	0.0467	18.0090
0.0002	26.6	2500	0.0468	18.2155
0.0002	27.13	2550	0.0471	18.1639
0.0002	27.66	2600	0.0472	17.92
0.0002	28.19	2650	0.0472	17.9303
0.0002	28.72	2700	0.0474	17.8116
0.0002	29.26	2750	0.0476	17.9045
0.0002	29.79	2800	0.0477	17.4942
0.0002	30.32	2850	0.0477	17.6129
0.0002	30.85	2900	0.0479	17.3910
0.0002	31.38	2950	0.0480	17.6594
0.0001	31.91	3000	0.0480	17.5303
0.0001	32.45	3050	0.0481	17.4245
0.0001	32.98	3100	0.0483	17.4413
0.0001	33.51	3150	0.0483	17.4013
0.0001	34.04	3200	0.0484	17.3342
0.0001	34.57	3250	0.0485	17.2361
0.0001	35.11	3300	0.0486	17.3613
0.0001	35.64	3350	0.0487	17.2606
0.0001	36.17	3400	0.0488	17.4039
0.0001	36.7	3450	0.0488	17.2168
0.0001	37.23	3500	0.0489	17.2194
0.0001	37.77	3550	0.0488	17.3032
0.0001	38.3	3600	0.0489	17.3303
0.0001	38.83	3650	0.0490	17.3277
0.0001	39.36	3700	0.0490	17.3381
0.0001	39.89	3750	0.0491	17.3471
0.0001	40.43	3800	0.0492	17.3497
0.0001	40.96	3850	0.0492	17.3484
0.0001	41.49	3900	0.0493	17.3910
0.0001	42.02	3950	0.0491	17.3019
0.0	42.55	4000	0.0492	17.2942
0.0	43.09	4050	0.0493	17.2645
0.0	43.62	4100	0.0493	17.2387
0.0	44.15	4150	0.0493	17.2348
0.0	44.68	4200	0.0493	17.2490
0.0	45.21	4250	0.0494	17.2374
0.0	45.74	4300	0.0495	17.2568
0.0	46.28	4350	0.0495	17.2619
0.0	46.81	4400	0.0495	17.2310
0.0	47.34	4450	0.0496	17.2374
0.0	47.87	4500	0.0496	17.2426
0.0	48.4	4550	0.0496	17.2387
0.0	48.94	4600	0.0496	17.2335
0.0	49.47	4650	0.0496	17.2387
0.0	50.0	4700	0.0496	17.2413

Framework versions

• Transformers 4.35.2
• Pytorch 2.0.0+cpu
• Datasets 2.15.0
• Tokenizers 0.15.0