snap-plate-stack-instance-dfine-v5

This model is a fine-tuned version of aguxez/snap-plate-instance-localization-dfine-v4 on the aguxez/snap-plate-stack-instance-v5 dataset. It achieves the following results on the evaluation set:

• Loss: 3.5716
• Map: 0.8835
• Map 50: 1.0
• Map 75: 1.0
• Map Small: -1.0
• Map Medium: 0.9
• Map Large: 0.8831
• Mar 1: 0.3375
• Mar 10: 0.9375
• Mar 100: 0.95
• Mar Small: -1.0
• Mar Medium: 0.9
• Mar Large: 0.9571
• Map Weight Plate: 0.8835
• Mar 100 Weight Plate: 0.95

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-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 7
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 0.05
• num_epochs: 8.0
• 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 Weight Plate	Mar 100 Weight Plate
19.4634	1.0	7	2.7761	0.8851	1.0	1.0	-1.0	1.0	0.8799	0.325	0.925	0.975	-1.0	1.0	0.9714	0.8851	0.975
21.5298	2.0	14	3.0125	0.8701	1.0	1.0	-1.0	0.9	0.869	0.325	0.9125	0.9375	-1.0	0.9	0.9429	0.8701	0.9375
17.9227	3.0	21	3.1628	0.8969	1.0	1.0	-1.0	1.0	0.8855	0.3375	0.9625	0.975	-1.0	1.0	0.9714	0.8969	0.975
20.3749	4.0	28	3.2891	0.8691	0.9743	0.9743	-1.0	1.0	0.8843	0.35	0.95	0.9625	-1.0	1.0	0.9571	0.8691	0.9625
17.0228	5.0	35	3.3509	0.8836	1.0	1.0	-1.0	0.9	0.8844	0.3375	0.9375	0.95	-1.0	0.9	0.9571	0.8836	0.95
19.0717	6.0	42	3.4911	0.8787	1.0	1.0	-1.0	0.9	0.8791	0.325	0.925	0.925	-1.0	0.9	0.9286	0.8787	0.925
16.4956	7.0	49	3.5982	0.8778	1.0	1.0	-1.0	0.9	0.8756	0.325	0.925	0.925	-1.0	0.9	0.9286	0.8778	0.925
18.0839	8.0	56	3.5716	0.8835	1.0	1.0	-1.0	0.9	0.8831	0.3375	0.9375	0.95	-1.0	0.9	0.9571	0.8835	0.95

Framework versions

• Transformers 5.14.1
• Pytorch 2.13.0+cu130
• Datasets 5.0.1
• Tokenizers 0.22.2