GenerativeImage2Text-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: 0.0544
• Wer Score: 2.6810

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: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Score
7.2829	1.8182	50	4.4333	11.5086
2.2118	3.6364	100	0.3655	1.1034
0.1089	5.4545	150	0.0428	1.0259
0.0223	7.2727	200	0.0421	0.4655
0.0162	9.0909	250	0.0430	0.4224
0.0139	10.9091	300	0.0434	0.9569
0.0126	12.7273	350	0.0455	0.8534
0.0115	14.5455	400	0.0457	3.2845
0.0106	16.3636	450	0.0490	2.3190
0.0096	18.1818	500	0.0515	2.2241
0.0094	20.0	550	0.0520	1.9569
0.0087	21.8182	600	0.0540	4.2328
0.0084	23.6364	650	0.0539	1.8448
0.008	25.4545	700	0.0546	2.5431
0.0077	27.2727	750	0.0542	2.7672
0.0072	29.0909	800	0.0544	2.6810

Framework versions

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