TRuCAL / tests /test_fixed_layer.py

Upload 24 files

e2847fd verified about 1 year ago

1.93 kB

import torch
from components.tiny_confessional_layer_fixed import TinyConfessionalLayer

def test_fixed_layer():
    """Test the fixed TinyConfessionalLayer with various input shapes."""
    print("🧪 Testing TinyConfessionalLayer with shape safety...")
    
    # Test cases with different input shapes
    test_cases = [
        (1, 10, 256),    # Standard input
        (2, 8, 512),     # Different batch/seq
        (4, 20, 128),    # Different dimensions
        (1, 5, 768),     # Larger feature dimension
        (3, 3, 3),       # Very small dimensions
    ]
    
    for batch, seq, d_model in test_cases:
        print(f"\nTesting: batch={batch}, seq={seq}, d_model={d_model}")
        
        try:
            # Create model with default d_model=256
            model = TinyConfessionalLayer(
                d_model=256,  # Fixed internal dimension
                enable_ambient=False,  # Disable ambient for simpler testing
                enable_windsurf=False  # Disable windsurf for now
            )
            
            # Create random input
            x = torch.randn(batch, seq, d_model)
            
            # Run forward pass
            out, metadata = model(x, audit_mode=True)
            
            # Check output shape
            expected_shape = (batch, seq, 256)  # Should match model's d_model
            assert out.shape == expected_shape, \
                f"Expected shape {expected_shape}, got {out.shape}"
                
            print(f"✅ Success! Input: {x.shape} -> Output: {out.shape}")
            print(f"   Cycles: {metadata['cycles_run']}, "
                  f"Shape fixes: {metadata.get('shape_issues_resolved', 0)}")
            
        except Exception as e:
            print(f"❌ Test failed: {str(e)}")
            import traceback
            traceback.print_exc()

if __name__ == "__main__":
    test_fixed_layer()
    print("\n🎉 All tests completed!")

	import torch
	from components.tiny_confessional_layer_fixed import TinyConfessionalLayer

	def test_fixed_layer():
	"""Test the fixed TinyConfessionalLayer with various input shapes."""
	print("🧪 Testing TinyConfessionalLayer with shape safety...")

	# Test cases with different input shapes
	test_cases = [
	(1, 10, 256), # Standard input
	(2, 8, 512), # Different batch/seq
	(4, 20, 128), # Different dimensions
	(1, 5, 768), # Larger feature dimension
	(3, 3, 3), # Very small dimensions
	]

	for batch, seq, d_model in test_cases:
	print(f"\nTesting: batch={batch}, seq={seq}, d_model={d_model}")

	try:
	# Create model with default d_model=256
	model = TinyConfessionalLayer(
	d_model=256, # Fixed internal dimension
	enable_ambient=False, # Disable ambient for simpler testing
	enable_windsurf=False # Disable windsurf for now
	)

	# Create random input
	x = torch.randn(batch, seq, d_model)

	# Run forward pass
	out, metadata = model(x, audit_mode=True)

	# Check output shape
	expected_shape = (batch, seq, 256) # Should match model's d_model
	assert out.shape == expected_shape, \
	f"Expected shape {expected_shape}, got {out.shape}"

	print(f"✅ Success! Input: {x.shape} -> Output: {out.shape}")
	print(f" Cycles: {metadata['cycles_run']}, "
	f"Shape fixes: {metadata.get('shape_issues_resolved', 0)}")

	except Exception as e:
	print(f"❌ Test failed: {str(e)}")
	import traceback
	traceback.print_exc()

	if __name__ == "__main__":
	test_fixed_layer()
	print("\n🎉 All tests completed!")