git-base-pokemon

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

• Loss: 0.0260
• Wer Score: 2.5870

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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
7.3233	2.8571	50	4.4354	22.5705
2.2055	5.7143	100	0.3490	10.2262
0.1031	8.5714	150	0.0287	0.3700
0.0177	11.4286	200	0.0197	1.4176
0.007	14.2857	250	0.0214	4.7060
0.0029	17.1429	300	0.0232	3.8049
0.0017	20.0	350	0.0237	4.6676
0.0012	22.8571	400	0.0241	4.0458
0.0011	25.7143	450	0.0246	3.7335
0.001	28.5714	500	0.0249	3.2042
0.0011	31.4286	550	0.0253	2.9423
0.0011	34.2857	600	0.0257	2.7527
0.0011	37.1429	650	0.0256	2.7015
0.0011	40.0	700	0.0258	2.7152
0.001	42.8571	750	0.0260	2.5531
0.001	45.7143	800	0.0260	2.6126
0.001	48.5714	850	0.0260	2.5870

Framework versions

• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1