segformer-b1-GFB-exp

This model is a fine-tuned version of nvidia/mit-b1 on the segments/GFB dataset. It achieves the following results on the evaluation set:

• eval_loss: 0.3306
• eval_mean_iou: 0.6108
• eval_mean_accuracy: 0.7335
• eval_overall_accuracy: 0.8871
• eval_accuracy_unlabeled: 0.9407
• eval_accuracy_GBM: 0.7830
• eval_accuracy_Podo: 0.7113
• eval_accuracy_Endo: 0.4989
• eval_iou_unlabeled: 0.8871
• eval_iou_GBM: 0.6348
• eval_iou_Podo: 0.5239
• eval_iou_Endo: 0.3975
• eval_runtime: 16.627
• eval_samples_per_second: 20.389
• eval_steps_per_second: 1.323
• epoch: 12.9412
• step: 1100

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• 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: cosine
• lr_scheduler_warmup_steps: 250
• num_epochs: 50

Framework versions

• Transformers 4.57.1
• Pytorch 2.9.1+cu130
• Datasets 4.4.1
• Tokenizers 0.22.1