finetune-instance-segmentation-ade20k-mini-mask2former

This model is a fine-tuned version of facebook/mask2former-swin-tiny-coco-instance on the qubvel-hf/ade20k-mini dataset. It achieves the following results on the evaluation set:

• Loss: 31.0067
• Map: 0.2061
• Map 50: 0.4076
• Map 75: 0.1946
• Map Small: 0.1368
• Map Medium: 0.623
• Map Large: 0.82
• Mar 1: 0.0921
• Mar 10: 0.2488
• Mar 100: 0.2856
• Mar Small: 0.2105
• Mar Medium: 0.7164
• Mar Large: 0.8705
• Map Person: 0.1413
• Mar 100 Person: 0.2035
• Map Car: 0.2709
• Mar 100 Car: 0.3678

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 2.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Map	Map 50	Map 75	Map Small	Map Medium	Map Large	Mar 1	Mar 10	Mar 100	Mar Small	Mar Medium	Mar Large	Map Person	Mar 100 Person	Map Car	Mar 100 Car
37.3934	1.0	100	33.6332	0.1981	0.3956	0.1799	0.1285	0.6173	0.7992	0.0896	0.2453	0.2821	0.2074	0.7135	0.8354	0.1349	0.2001	0.2613	0.3641
29.0441	2.0	200	31.0067	0.2061	0.4076	0.1946	0.1368	0.623	0.82	0.0921	0.2488	0.2856	0.2105	0.7164	0.8705	0.1413	0.2035	0.2709	0.3678

Framework versions

• Transformers 4.48.0.dev0
• Pytorch 2.5.0+cu121
• Datasets 2.19.1
• Tokenizers 0.21.0