paligemma_clevr

This model is a fine-tuned version of leo009/paligemma-3b-pt-224 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4210

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: 2e-05
• train_batch_size: 3
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 12
• optimizer: Use adamw_hf with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss
1.463	0.1715	100	0.5191
0.5064	0.3429	200	0.4908
0.4851	0.5144	300	0.4699
0.4797	0.6858	400	0.4539
0.4608	0.8573	500	0.4486
0.4416	1.0274	600	0.4425
0.4253	1.1989	700	0.4316
0.4015	1.3703	800	0.4245
0.3976	1.5418	900	0.4262
0.4092	1.7132	1000	0.4193
0.399	1.8847	1100	0.4210

Framework versions

• PEFT 0.14.0
• Transformers 4.48.0
• Pytorch 2.4.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0