paligemma_racer_fp32

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

• Loss: 2.5338

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: 16
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• 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: 1000
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss
3.942	0.1674	100	3.2672
2.8843	0.3349	200	2.7924
2.675	0.5023	300	2.6303
2.5428	0.6697	400	2.5909
2.5034	0.8372	500	2.5196
2.4216	1.0046	600	2.5330
2.377	1.1720	700	2.5048
2.3428	1.3395	800	2.5692
2.3453	1.5069	900	2.5152
2.3258	1.6743	1000	2.5434
2.2206	1.8418	1100	2.5338

Framework versions

• Transformers 4.46.3
• Pytorch 2.5.1
• Datasets 3.1.0
• Tokenizers 0.20.3