================================================================================
Model Configuration (CRITICAL FOR INFERENCE)
================================================================================
Task: classification
Genome: [[[9, 1], [5, 1]], [[8, 1], [2, 1]], [[9, 1], [9, 1]], [[9, 4], [8, 1]], [[7, 3], [3, 2]]]
Architecture: AdapterGenotype(ops=[('spatial_attention', 1), ('dil_conv_3x3', 1), ('channel_attention', 1), ('conv_1x1', 1), ('spatial_attention', 1), ('spatial_attention', 1), ('spatial_attention', 4), ('channel_attention', 1), ('max_pool_3x3', 3), ('sep_conv_3x3', 2)], ops_concat=[5, 6])
backbone: yolov8n
search_space: micro
num_cells: 2
num_classes: 10
image_size: 224
================================================================================
Performance Metrics
================================================================================
Params: 1.94955 MB
Forward FLOPs: 229.249040 M
Training FLOPs: 386.895952 M [thop]
Training Efficiency: 43.7% savings
--------------------------------------------------------------------------------
Epoch  | Train Acc  | Valid Acc  | Error      | FLOPs     
--------------------------------------------------------------------------------
0      | 57.93      | 75.14      | 24.86      | 386.90    
1      | 65.15      | 80.04      | 19.96      | 386.90    
