paligemma_newslakeandmadvqa_conbime

This model is a fine-tuned version of google/paligemma-3b-pt-224 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.0349

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: 10
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 40
• optimizer: Use OptimizerNames.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: 900
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
5.6592	0.1768	100	5.9832
5.1792	0.3535	200	4.5564
3.6303	0.5303	300	3.2205
2.8756	0.7070	400	2.5592
2.3211	0.8838	500	2.1059
1.9172	1.0605	600	1.8195
1.7352	1.2373	700	1.6310
1.621	1.4141	800	1.4883
1.4855	1.5908	900	1.3832
1.4496	1.7676	1000	1.2959
1.2769	1.9443	1100	1.2350
1.2199	2.1211	1200	1.1831
1.2619	2.2978	1300	1.1391
1.1412	2.4746	1400	1.0943
1.0869	2.6513	1500	1.0589
1.123	2.8281	1600	1.0349

Framework versions

• PEFT 0.13.0
• Transformers 4.46.0.dev0
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0