vision-tags-paligemma-qlora

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

• Loss: 0.2083

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: 32
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• 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: 2
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
1.7663	0.0580	100	0.3039
1.0343	0.1159	200	0.2727
1.0763	0.1739	300	0.2598
0.9531	0.2318	400	0.2480
0.8547	0.2898	500	0.2407
0.9085	0.3477	600	0.2314
0.8865	0.4057	700	0.2262
0.7912	0.4636	800	0.2216
0.8547	0.5216	900	0.2178
0.8279	0.5795	1000	0.2147
0.8697	0.6375	1100	0.2129
0.8538	0.6955	1200	0.2104
0.8126	0.7534	1300	0.2098
0.8165	0.8114	1400	0.2089
0.8485	0.8693	1500	0.2086
0.8075	0.9273	1600	0.2083
0.8574	0.9852	1700	0.2083

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1