Image Classification Results

Config & Metrics
Train/Validation Plots
Test Plots

Training Setup

ParameterValue
Model Nameresnet18
EpochsBecause of early stopping: the trainingstopped at epoch 7
Batch SizeAuto-selected batch size by Ludwig:
16
Fine TuneTrue
Use PretrainedTrue
Learning RateAuto-selected learning rate by Ludwig:
1e-05
Based on model architecture and training setup (e.g., fine-tuning).
See Ludwig Trainer Parameters for details.
Random Seedearly StopN/A
Data SplitDetected a split column (with values 0 and 2) in the input CSV. Used this column as a base andreassigned 15.0% of the training set (originally labeled 0) to validation (labeled 1).

Model trained using Ludwig.
If want to learn more about Ludwig default settings,please check the their website(ludwig.ai).


Model Performance Summary

MetricTrainValidationTest
accuracy0.84170.15000.2000
accuracy_micro0.84710.20000.2000
hits_at_k0.92500.45000.3000
loss0.67492.79072.8261
roc_auc0.99980.78240.6917

Training & Validation Visualizations

Learning Curves Label Accuracy

Learning Curves Label Hits At K

Learning Curves Label Loss

Test Visualizations

Compare Classifiers Multiclass Multimetric Label Best10

Compare Classifiers Multiclass Multimetric Label Sorted

Compare Classifiers Multiclass Multimetric Label Top10

Compare Classifiers Multiclass Multimetric Label Worst10

Compare Classifiers Performance From Prob

Compare Performance Label

Confusion Matrix Label Top10

Confusion Matrix Entropy Label Top10

Frequency Vs F1 Label

Roc Curves