git-large-r-coco-IDB2-V1

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

• Loss: 2.7997

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: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
23.592	2.5	5	10.2322
19.1654	5.0	10	8.4633
16.2373	7.5	15	7.4093
14.5374	10.0	20	6.7997
13.42	12.5	25	6.3244
12.5023	15.0	30	5.8997
11.6732	17.5	35	5.5051
10.9066	20.0	40	5.1363
10.1916	22.5	45	4.7917
9.5274	25.0	50	4.4732
8.9143	27.5	55	4.1802
8.3521	30.0	60	3.9132
7.843	32.5	65	3.6740
7.3891	35.0	70	3.4631
6.9916	37.5	75	3.2809
6.6506	40.0	80	3.1274
6.3665	42.5	85	3.0022
6.14	45.0	90	2.9061
5.9706	47.5	95	2.8389
5.8582	50.0	100	2.7997

Framework versions

• Transformers 4.46.1
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.20.2