maximuspowers/muat-mean-std-medium-classifier

Pattern Classifier

This model was trained to classify which patterns a subject model was trained on, based on neuron activation signatures.

Dataset

• Training Dataset: maximuspowers/muat-mean-std-medium
• Input Mode: signature
• Number of Patterns: 14

Patterns

The model predicts which of the following 14 patterns the subject model was trained to classify as positive:

1. palindrome
2. sorted_ascending
3. sorted_descending
4. alternating
5. contains_abc
6. starts_with
7. ends_with
8. no_repeats
9. has_majority
10. increasing_pairs
11. decreasing_pairs
12. vowel_consonant
13. first_last_match
14. mountain_pattern

Model Architecture

• Signature Encoder: [512, 256, 256, 128]
• Activation: relu
• Dropout: 0.2
• Batch Normalization: True

Training Configuration

• Optimizer: adam
• Learning Rate: 0.001
• Batch Size: 16
• Loss Function: BCE with Logits (with pos_weight for training, unweighted for validation)

Test Set Performance

• F1 Macro: 0.1749
• F1 Micro: 0.1751
• Hamming Accuracy: 0.7321
• Exact Match Accuracy: 0.0110
• BCE Loss: 0.5217

Per-Pattern Performance (Test Set)

Pattern	Precision	Recall	F1 Score
palindrome	9.5%	97.4%	17.4%
sorted_ascending	41.7%	10.5%	16.8%
sorted_descending	14.2%	18.7%	16.1%
alternating	18.9%	55.1%	28.2%
contains_abc	19.2%	22.5%	20.7%
starts_with	6.3%	58.0%	11.4%
ends_with	10.3%	79.8%	18.2%
no_repeats	6.2%	2.2%	3.2%
has_majority	36.8%	26.9%	31.1%
increasing_pairs	8.8%	5.1%	6.5%
decreasing_pairs	20.9%	24.7%	22.6%
vowel_consonant	25.0%	33.3%	28.6%
first_last_match	9.0%	94.7%	16.5%
mountain_pattern	9.4%	6.4%	7.6%