File size: 9,938 Bytes
4ac02ee
 
 
4f8aa53
4ac02ee
 
 
4f8aa53
 
 
 
 
 
 
 
 
 
 
4ac02ee
 
4f8aa53
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ac02ee
4f8aa53
4ac02ee
4f8aa53
4ac02ee
 
4f8aa53
 
4ac02ee
 
 
 
 
 
4f8aa53
 
4ac02ee
 
 
4f8aa53
4ac02ee
 
 
4f8aa53
 
4ac02ee
 
 
4f8aa53
4ac02ee
 
 
 
4f8aa53
 
4ac02ee
4f8aa53
4ac02ee
 
4f8aa53
 
 
 
 
 
 
 
 
4ac02ee
 
4f8aa53
 
 
 
 
 
 
 
4ac02ee
4f8aa53
4ac02ee
 
 
4f8aa53
 
4ac02ee
 
 
4f8aa53
 
 
4ac02ee
 
 
4f8aa53
4ac02ee
 
 
4f8aa53
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ac02ee
 
 
4f8aa53
 
 
 
 
 
 
 
 
 
 
4ac02ee
4f8aa53
4ac02ee
 
 
4f8aa53
 
4ac02ee
4f8aa53
 
 
 
 
 
 
 
 
 
 
 
4ac02ee
4f8aa53
4ac02ee
 
4f8aa53
4ac02ee
4f8aa53
 
 
 
 
4ac02ee
4f8aa53
 
4ac02ee
 
4f8aa53
 
 
 
 
 
 
 
 
 
4ac02ee
 
 
4f8aa53
 
 
 
 
 
 
 
 
 
4ac02ee
 
 
4f8aa53
 
 
4ac02ee
4f8aa53
4ac02ee
4f8aa53
 
4ac02ee
 
 
 
 
4f8aa53
 
4ac02ee
 
 
4f8aa53
 
 
4ac02ee
 
4f8aa53
 
 
 
 
 
 
 
 
 
 
 
 
4ac02ee
4f8aa53
 
 
4ac02ee
4f8aa53
 
 
 
 
 
 
 
 
 
4ac02ee
4f8aa53
 
 
 
 
4ac02ee
 
 
 
 
4f8aa53
 
4ac02ee
4f8aa53
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ac02ee
 
4f8aa53
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
import gradio as gr
import torch
import yaml
import os
from core.graph_mamba import GraphMamba
from data.loader import GraphDataLoader
from utils.metrics import GraphMetrics
from utils.visualization import GraphVisualizer
import warnings
warnings.filterwarnings('ignore')

# Force CPU for HuggingFace Spaces
if os.getenv('SPACE_ID') or os.getenv('GRADIO_SERVER_NAME'):
    device = torch.device('cpu')
    print("Running on HuggingFace Spaces - using CPU")
else:
    device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
    print(f"Running locally - using {device}")

# Load configuration
config = {
    'model': {
        'd_model': 128,  # Smaller for demo
        'd_state': 8,
        'd_conv': 4,
        'expand': 2,
        'n_layers': 3,  # Fewer layers for speed
        'dropout': 0.1
    },
    'data': {
        'batch_size': 16,
        'test_split': 0.2
    },
    'ordering': {
        'strategy': 'bfs',
        'preserve_locality': True
    }
}

# Global model holder
model = None
current_dataset = None

def load_and_evaluate(dataset_name, ordering_strategy, num_layers):
    """Load dataset, configure model, return results"""
    global model, config, current_dataset
    
    try:
        # Update config
        config['ordering']['strategy'] = ordering_strategy
        config['model']['n_layers'] = num_layers
        
        print(f"Loading dataset: {dataset_name}")
        
        # Load data
        data_loader = GraphDataLoader()
        
        if dataset_name in ['Cora', 'CiteSeer', 'PubMed']:
            dataset = data_loader.load_node_classification_data(dataset_name)
            data = dataset[0].to(device)
            task_type = 'node_classification'
            current_dataset = data
            print(f"Loaded {dataset_name}: {data.num_nodes} nodes, {data.num_edges} edges")
        else:
            dataset = data_loader.load_graph_classification_data(dataset_name)
            task_type = 'graph_classification'
            print(f"Loaded {dataset_name}: {len(dataset)} graphs")
        
        # Get dataset info
        dataset_info = data_loader.get_dataset_info(dataset)
        
        print(f"Dataset info: {dataset_info}")
        
        # Initialize model
        print("Initializing GraphMamba model...")
        model = GraphMamba(config).to(device)
        
        # Initialize classifier for evaluation
        num_classes = dataset_info['num_classes']
        model._init_classifier(num_classes, device)
        
        total_params = sum(p.numel() for p in model.parameters())
        print(f"Model parameters: {total_params:,}")
        
        # Quick evaluation (random weights for demo)
        print("Running evaluation...")
        if task_type == 'node_classification':
            # Use test mask for evaluation
            if hasattr(data, 'test_mask'):
                mask = data.test_mask
            else:
                # Create a test mask if not available
                num_nodes = data.num_nodes
                mask = torch.zeros(num_nodes, dtype=torch.bool)
                mask[num_nodes//2:] = True
            
            metrics = GraphMetrics.evaluate_node_classification(
                model, data, mask, device
            )
            
            # Create visualization
            print("Creating visualization...")
            fig = GraphVisualizer.create_graph_plot(data)
            
        else:
            # Graph classification
            train_loader, val_loader, test_loader = data_loader.create_dataloaders(
                dataset, 'graph_classification'
            )
            metrics = GraphMetrics.evaluate_graph_classification(
                model, test_loader, device
            )
            fig = GraphVisualizer.create_metrics_plot(metrics)
        
        # Format results
        results_text = f"""
## ๐Ÿง  Mamba Graph Neural Network Results

### Dataset: {dataset_name}

**Dataset Statistics:**
- ๐Ÿ“Š Features: {dataset_info['num_features']}
- ๐Ÿท๏ธ Classes: {dataset_info['num_classes']}
- ๐Ÿ“ˆ Graphs: {dataset_info['num_graphs']}
- ๐Ÿ”— Avg Nodes: {dataset_info['avg_nodes']:.1f}
- ๐ŸŒ Avg Edges: {dataset_info['avg_edges']:.1f}

**Model Configuration:**
- ๐Ÿ”„ Ordering Strategy: {ordering_strategy}
- ๐Ÿ—๏ธ Layers: {num_layers}
- โš™๏ธ Parameters: {sum(p.numel() for p in model.parameters()):,}
- ๐Ÿ’พ Device: {device}

**Performance Metrics:**
        """
        
        for metric, value in metrics.items():
            if isinstance(value, float) and metric != 'error':
                results_text += f"- ๐Ÿ“ˆ {metric.replace('_', ' ').title()}: {value:.4f}\n"
            elif metric == 'error':
                results_text += f"- โš ๏ธ Error: {value}\n"
        
        results_text += f"""

**Status:** โœ… Model successfully loaded and evaluated!

*Note: This is a demo with random weights. In production, the model would be trained on the dataset.*
        """
        
        print("Evaluation completed successfully!")
        return results_text, fig
        
    except Exception as e:
        error_msg = f"""
## โŒ Error Loading Model

**Error:** {str(e)}

**Troubleshooting:**
- Check dataset availability
- Verify device compatibility  
- Try different ordering strategy

**Debug Info:**
- Device: {device}
- Dataset: {dataset_name}
- Strategy: {ordering_strategy}
        """
        
        print(f"Error: {e}")
        
        # Return empty plot on error
        import plotly.graph_objects as go
        fig = go.Figure()
        fig.add_annotation(
            text=f"Error: {str(e)}",
            x=0.5, y=0.5,
            xref="paper", yref="paper",
            showarrow=False
        )
        
        return error_msg, fig

# Gradio interface
with gr.Blocks(
    title="๐Ÿง  Mamba Graph Neural Network",
    theme=gr.themes.Soft(),
    css="""
    .gradio-container {
        max-width: 1200px !important;
    }
    """
) as demo:
    
    gr.Markdown("""
    # ๐Ÿง  Mamba Graph Neural Network
    
    **Real-time evaluation of Graph-Mamba on standard benchmarks.**
    
    This demonstrates the revolutionary combination of Mamba's linear complexity with graph neural networks.
    Uses actual datasets and real model architectures - no synthetic data.
    
    ๐Ÿš€ **Features:**
    - Linear O(n) complexity for massive graphs
    - Multiple graph ordering strategies  
    - Real benchmark datasets (Cora, CiteSeer, etc.)
    - Interactive visualizations
    """)
    
    with gr.Row():
        with gr.Column(scale=1):
            gr.Markdown("### ๐ŸŽฎ Model Configuration")
            
            dataset_choice = gr.Dropdown(
                choices=['Cora', 'CiteSeer', 'PubMed', 'MUTAG', 'ENZYMES'],
                value='Cora',
                label="๐Ÿ“Š Dataset",
                info="Choose a graph dataset for evaluation"
            )
            
            ordering_choice = gr.Dropdown(
                choices=['bfs', 'spectral', 'degree', 'community'],
                value='bfs',
                label="๐Ÿ”„ Graph Ordering Strategy",
                info="How to convert graph to sequence"
            )
            
            layers_slider = gr.Slider(
                minimum=2, maximum=6, value=3, step=1,
                label="๐Ÿ—๏ธ Number of Mamba Layers",
                info="More layers = more capacity"
            )
            
            evaluate_btn = gr.Button(
                "๐Ÿš€ Evaluate Model", 
                variant="primary",
                size="lg"
            )
            
            gr.Markdown("""
            ### ๐Ÿ“– Ordering Strategies:
            - **BFS**: Breadth-first traversal
            - **Spectral**: Eigenvalue-based ordering  
            - **Degree**: High-degree nodes first
            - **Community**: Cluster-aware ordering
            """)
        
        with gr.Column(scale=2):
            results_text = gr.Markdown("""
            ### ๐Ÿ‘‹ Welcome!
            
            Select your parameters and click **'๐Ÿš€ Evaluate Model'** to see Mamba Graph in action.
            
            The model will:
            1. ๐Ÿ“ฅ Load the selected dataset
            2. ๐Ÿ”„ Apply graph ordering strategy  
            3. ๐Ÿง  Process through Mamba layers
            4. ๐Ÿ“Š Evaluate performance
            5. ๐Ÿ“ˆ Show results and visualization
            """)
    
    with gr.Row():
        with gr.Column():
            visualization = gr.Plot(
                label="๐Ÿ“ˆ Graph Visualization",
                container=True
            )
    
    # Event handlers
    evaluate_btn.click(
        fn=load_and_evaluate,
        inputs=[dataset_choice, ordering_choice, layers_slider],
        outputs=[results_text, visualization],
        show_progress=True
    )
    
    # Example section
    gr.Markdown("""
    ---
    ### ๐ŸŽฏ What Makes This Special?
    
    **Traditional GNNs:** O(nยฒ) complexity limits them to small graphs
    
    **Mamba Graph:** O(n) complexity enables processing of massive graphs
    
    **Key Innovation:** Smart graph-to-sequence conversion preserves structural information while enabling linear-time processing.
    
    ### ๐Ÿ”ฌ Technical Details:
    - **Selective State Space Models** for sequence processing
    - **Structure-preserving ordering** algorithms
    - **Position encoding** to maintain graph relationships
    - **Multi-scale processing** for different graph properties
    
    ### ๐Ÿ“š References:
    - Mamba: Linear-Time Sequence Modeling (Gu & Dao, 2023)
    - Graph Neural Networks (Kipf & Welling, 2017)
    - Spectral Graph Theory applications
    """)

if __name__ == "__main__":
    print("๐Ÿง  Starting Mamba Graph Demo...")
    print(f"Device: {device}")
    print("Loading Gradio interface...")
    
    demo.launch(
        server_name="0.0.0.0",
        server_port=7860,
        show_error=True,
        share=False  # Set to False for HuggingFace Spaces
    )