resultsroco__captions__google_paligemma-3b-mix-448__fullrun__2911-095336

This model is a fine-tuned version of google/paligemma-3b-mix-448 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.6451

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_steps: 2
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.8798	0.9998	1022	2.8440
2.7474	1.9996	2044	2.7501
2.6883	2.9994	3066	2.7063
2.6667	3.9993	4088	2.6809
2.5908	4.9991	5110	2.6667
2.6089	5.9999	6133	2.6562
2.5774	6.9997	7155	2.6467
2.5092	7.9995	8177	2.6429
2.5263	8.9993	9199	2.6425
2.4526	9.9982	10220	2.6451

Framework versions

• PEFT 0.13.0
• Transformers 4.45.1
• Pytorch 2.3.0.post101
• Datasets 2.19.1
• Tokenizers 0.20.0