paligemma_VQAv2_ko

This model is a fine-tuned version of google/paligemma-3b-pt-224 on an korean VQAv2 dataset.(HuggingFaceM4/VQAv2)

ds = load_dataset('HuggingFaceM4/VQAv2', split="train[:100%]",cache_dir=root_path+'dataset/')

Model description

This model was finetuned using VQAv2 datasets translated into KOREAN.

Intended uses & limitations

Nothing

Training and evaluation data

Nothing

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 1
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2
• num_epochs: 2

Framework versions

• PEFT 0.8.2
• Transformers 4.45.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1