spaceom-3b

This model is a fine-tuned version of UCSC-VLAA/VLAA-Thinker-Qwen2.5VL-3B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6695

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: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
0.7544	0.2384	500	0.7171
0.7013	0.4768	1000	0.7022
0.6639	0.7153	1500	0.6930
0.6997	0.9537	2000	0.6865
0.6678	1.1917	2500	0.6827
0.6852	1.4301	3000	0.6797
0.6852	1.6685	3500	0.6766
0.6645	1.9070	4000	0.6748
0.686	2.1450	4500	0.6736
0.6444	2.3834	5000	0.6721
0.6697	2.6218	5500	0.6708
0.6684	2.8602	6000	0.6695

Framework versions

• PEFT 0.15.2
• Transformers 4.51.0
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1