Hugging Face's logo

NICOPOI-9
/
segformer-b3-finetuned-morphpadver1-hgo-coord-v4

Image Segmentation
Transformers
Safetensors
segformer
vision
Generated from Trainer
Model card Files Community
segformer-b3-finetuned-morphpadver1-hgo-coord-v4 / README.md
NICOPOI-9's picture
NICOPOI-9
End of training
ffc7027 verified
preview code
|
raw
history blame contribute delete
5.83 kB
metadata
library_name: transformers
license: other
base_model: nvidia/mit-b3
tags:
  - vision
  - image-segmentation
  - generated_from_trainer
model-index:
  - name: segformer-b3-finetuned-morphpadver1-hgo-coord-v4
    results: []

segformer-b3-finetuned-morphpadver1-hgo-coord-v4

This model is a fine-tuned version of nvidia/mit-b3 on the NICOPOI-9/morphpad_coord_hgo_512_4class_v2 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0105
  • Mean Iou: 0.9978
  • Mean Accuracy: 0.9989
  • Overall Accuracy: 0.9989
  • Accuracy 0-0: 0.9991
  • Accuracy 0-90: 0.9988
  • Accuracy 90-0: 0.9988
  • Accuracy 90-90: 0.9989
  • Iou 0-0: 0.9983
  • Iou 0-90: 0.9977
  • Iou 90-0: 0.9978
  • Iou 90-90: 0.9974

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: 6e-05
  • train_batch_size: 1
  • eval_batch_size: 1
  • seed: 42
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • num_epochs: 50

Training results

Training Loss Epoch Step Validation Loss Mean Iou Mean Accuracy Overall Accuracy Accuracy 0-0 Accuracy 0-90 Accuracy 90-0 Accuracy 90-90 Iou 0-0 Iou 0-90 Iou 90-0 Iou 90-90
1.0526 2.6525 4000 1.0174 0.3366 0.5028 0.5035 0.5781 0.3249 0.6323 0.4758 0.3780 0.2714 0.3136 0.3836
0.7543 5.3050 8000 0.5721 0.5952 0.7459 0.7458 0.7487 0.7334 0.6962 0.8051 0.6389 0.5638 0.5555 0.6227
0.1714 7.9576 12000 0.1845 0.8808 0.9366 0.9366 0.9416 0.9454 0.9140 0.9455 0.8984 0.8681 0.8613 0.8956
0.2092 10.6101 16000 0.0953 0.9378 0.9679 0.9679 0.9719 0.9663 0.9730 0.9602 0.9466 0.9322 0.9312 0.9412
0.0773 13.2626 20000 0.1115 0.9348 0.9663 0.9663 0.9665 0.9728 0.9588 0.9672 0.9450 0.9220 0.9319 0.9405
0.1 15.9151 24000 0.0842 0.9597 0.9794 0.9794 0.9792 0.9743 0.9802 0.9840 0.9642 0.9559 0.9539 0.9649
0.0297 18.5676 28000 0.0524 0.9706 0.9851 0.9851 0.9855 0.9868 0.9886 0.9794 0.9753 0.9676 0.9666 0.9728
0.0188 21.2202 32000 0.0463 0.9813 0.9906 0.9906 0.9911 0.9884 0.9899 0.9930 0.9828 0.9804 0.9787 0.9834
0.0987 23.8727 36000 0.0377 0.9849 0.9924 0.9924 0.9913 0.9967 0.9885 0.9932 0.9866 0.9816 0.9844 0.9871
0.1753 26.5252 40000 0.0367 0.9878 0.9939 0.9938 0.9922 0.9929 0.9947 0.9957 0.9876 0.9879 0.9853 0.9903
0.0536 29.1777 44000 0.0392 0.9880 0.9939 0.9939 0.9946 0.9945 0.9933 0.9934 0.9891 0.9875 0.9859 0.9893
0.0273 31.8302 48000 0.0450 0.9879 0.9939 0.9939 0.9946 0.9922 0.9937 0.9952 0.9897 0.9850 0.9872 0.9898
0.006 34.4828 52000 0.0272 0.9936 0.9968 0.9968 0.9972 0.9968 0.9966 0.9965 0.9940 0.9938 0.9927 0.9937
0.005 37.1353 56000 0.0240 0.9948 0.9974 0.9974 0.9981 0.9961 0.9975 0.9980 0.9965 0.9936 0.9951 0.9941
0.0039 39.7878 60000 0.0244 0.9954 0.9977 0.9977 0.9967 0.9981 0.9980 0.9981 0.9953 0.9949 0.9951 0.9965
0.0042 42.4403 64000 0.0203 0.9961 0.9980 0.9980 0.9982 0.9979 0.9982 0.9979 0.9972 0.9956 0.9954 0.9962
0.0034 45.0928 68000 0.0165 0.9970 0.9985 0.9985 0.9984 0.9984 0.9984 0.9987 0.9976 0.9966 0.9962 0.9975
0.0033 47.7454 72000 0.0105 0.9978 0.9989 0.9989 0.9991 0.9988 0.9988 0.9989 0.9983 0.9977 0.9978 0.9974

Framework versions

  • Transformers 4.48.3
  • Pytorch 2.1.0
  • Datasets 3.2.0
  • Tokenizers 0.21.0