git-base-naruto

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: 0.0613
• Wer Score: 4.6462

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
7.3247	3.7037	50	4.4756	6.1692
2.2782	7.4074	100	0.4117	0.4308
0.1182	11.1111	150	0.0433	0.4462
0.0162	14.8148	200	0.0483	0.5231
0.0105	18.5185	250	0.0527	0.5231
0.0085	22.2222	300	0.0548	0.4769
0.007	25.9259	350	0.0578	0.8923
0.006	29.6296	400	0.0599	0.8462
0.0051	33.3333	450	0.0598	6.0
0.004	37.0370	500	0.0608	5.5538
0.0035	40.7407	550	0.0606	7.7077
0.0028	44.4444	600	0.0611	5.4308
0.0023	48.1481	650	0.0613	4.6462

Framework versions

• Transformers 4.40.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1