Skip to content

Hyperparam optimization with augmented data #1 #12

Description

@hbandukw

DATA: AUGMENTED / BALANCED
METRIC: ACCURACY
HYPERPARAMS:
[I 2025-01-24 19:55:40,455] Trial 27 finished with value: 0.904019688269073 and parameters: {'learning_rate': 8.300432875328772e-05, 'num_trainable_layers': 2, 'dropout_rate': 0.3847406475130443, 'batch_size': 32, 'step_size': 9, 'gamma': 0.5951936405857416, 'epochs': 5}. Best is trial 27 with value: 0.904019688269073.

RESULTS: Epoch 5 - Train Loss: 0.6058, Val Loss: 1.1301, Val Accuracy: 0.6579
Final Model Accuracy: 0.6579

see comment below for differences in results

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

No labels
No labels

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions