File size: 12,754 Bytes
001f5b1
 
d4c04d4
 
001f5b1
 
 
 
4f57881
001f5b1
 
 
 
 
 
 
 
 
d4c04d4
001f5b1
 
756e842
d4c04d4
001f5b1
 
 
 
ceb55e4
001f5b1
 
 
 
d4c04d4
 
001f5b1
d4c04d4
 
 
001f5b1
 
 
d4c04d4
001f5b1
 
 
 
 
 
ceb55e4
d4c04d4
001f5b1
 
 
 
d4c04d4
 
 
 
001f5b1
 
 
 
d4c04d4
 
001f5b1
 
 
 
d4c04d4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ceb55e4
 
 
d4c04d4
f5b1522
d4c04d4
f5b1522
d4c04d4
 
 
 
 
 
 
 
f5b1522
d4c04d4
f5b1522
 
 
d4c04d4
 
f5b1522
001f5b1
 
 
 
 
 
d4c04d4
001f5b1
 
d4c04d4
 
001f5b1
 
 
 
ceb55e4
001f5b1
 
 
 
d4c04d4
001f5b1
 
 
 
 
 
d4c04d4
 
 
6aa1363
 
001f5b1
 
 
d4c04d4
001f5b1
 
d4c04d4
001f5b1
 
 
 
 
 
d4c04d4
 
 
001f5b1
 
 
 
 
 
 
d4c04d4
001f5b1
d4c04d4
001f5b1
 
 
d4c04d4
001f5b1
 
 
 
 
 
 
d4c04d4
 
 
 
001f5b1
d4c04d4
001f5b1
d4c04d4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6aa1363
001f5b1
d4c04d4
 
001f5b1
d4c04d4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
756e842
d4c04d4
 
 
 
001f5b1
d4c04d4
756e842
d4c04d4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
001f5b1
 
d4c04d4
 
 
 
 
 
 
001f5b1
 
 
ceb55e4
001f5b1
 
 
 
 
d4c04d4
6aa1363
d4c04d4
 
001f5b1
d4c04d4
001f5b1
 
d4c04d4
 
001f5b1
ceb55e4
 
d4c04d4
ceb55e4
d4c04d4
ceb55e4
d4c04d4
 
ceb55e4
 
d4c04d4
 
 
8c8b756
ceb55e4
8c8b756
 
d4c04d4
 
8c8b756
 
d4c04d4
c4555aa
001f5b1
d4c04d4
 
 
 
 
 
 
 
 
 
 
 
 
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
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
#!/usr/bin/env python3
"""

Flask App with Gunicorn for Deep Modal Files

Economics Chat Application using Qwen2 model

"""

from flask import Flask, request, jsonify, render_template_string
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
import os
import logging

# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

app = Flask(__name__)

# Global variables for model and tokenizer
model = None
tokenizer = None

# HTML template
HTML_TEMPLATE = """

<!DOCTYPE html>

<html>

<head>

    <title>AEGIS Economics AI</title>

    <meta charset="utf-8">

    <meta name="viewport" content="width=device-width, initial-scale=1">

    <style>

        body { font-family: Arial, sans-serif; margin: 0; padding: 20px; background: #f5f5f5; }

        .container { max-width: 800px; margin: 0 auto; background: white; padding: 20px; border-radius: 10px; box-shadow: 0 2px 10px rgba(0,0,0,0.1); }

        .header { text-align: center; margin-bottom: 30px; }

        .chat-container { border: 1px solid #ddd; border-radius: 5px; height: 400px; overflow-y: auto; padding: 10px; margin-bottom: 20px; background: #fafafa; }

        .message { margin: 10px 0; padding: 10px; border-radius: 5px; }

        .user-message { background: #007bff; color: white; margin-left: 20%; }

        .ai-message { background: #e9ecef; color: #333; margin-right: 20%; }

        .input-group { display: flex; gap: 10px; }

        .input-field { flex: 1; padding: 10px; border: 1px solid #ddd; border-radius: 5px; }

        .send-btn { padding: 10px 20px; background: #007bff; color: white; border: none; border-radius: 5px; cursor: pointer; }

        .send-btn:hover { background: #0056b3; }

        .loading { text-align: center; color: #666; font-style: italic; }

    </style>

</head>

<body>

    <div class="container">

        <div class="header">

            <h1>🏛️ AEGIS Economics AI</h1>

            <p>Advanced Economic Analysis & Policy Insights</p>

        </div>

        

        <div id="chat-container" class="chat-container">

            <div class="message ai-message">

                Hello! I'm AEGIS Economics AI. Ask me about economic policies, market analysis, or financial strategies.

                <div id="model-status" style="font-size: 0.8em; color: #666; margin-top: 5px;">

                    Checking model status...

                </div>

            </div>

        </div>

        

        <div class="input-group">

            <input type="text" id="user-input" class="input-field" placeholder="Ask about economics, policy, markets..." onkeypress="handleKeyPress(event)">

            <button onclick="sendMessage()" class="send-btn">Send</button>

        </div>

    </div>



    <script>

        // Check model status on page load

        async function checkModelStatus() {

            try {

                const response = await fetch('/health');

                const data = await response.json();

                const statusDiv = document.getElementById('model-status');

                

                if (data.model_loaded) {

                    statusDiv.textContent = '✅ Model loaded and ready!';

                    statusDiv.style.color = '#28a745';

                } else {

                    statusDiv.textContent = '⏳ Model loading... Please wait.';

                    statusDiv.style.color = '#ffc107';

                    // Try to load model

                    setTimeout(tryLoadModel, 2000);

                }

            } catch (error) {

                const statusDiv = document.getElementById('model-status');

                statusDiv.textContent = '❌ Connection error';

                statusDiv.style.color = '#dc3545';

            }

        }



        async function tryLoadModel() {

            try {

                const response = await fetch('/load_model', { method: 'POST' });

                const data = await response.json();

                

                if (data.success) {

                    const statusDiv = document.getElementById('model-status');

                    statusDiv.textContent = '✅ Model loaded successfully!';

                    statusDiv.style.color = '#28a745';

                } else {

                    setTimeout(checkModelStatus, 5000); // Check again in 5 seconds

                }

            } catch (error) {

                setTimeout(checkModelStatus, 5000);

            }

        }



        // Call on page load

        window.onload = checkModelStatus;



        function handleKeyPress(event) {

            if (event.key === 'Enter') {

                sendMessage();

            }

        }



        function addMessage(content, isUser) {

            const chatContainer = document.getElementById('chat-container');

            const messageDiv = document.createElement('div');

            messageDiv.className = `message ${isUser ? 'user-message' : 'ai-message'}`;

            messageDiv.textContent = content;

            chatContainer.appendChild(messageDiv);

            chatContainer.scrollTop = chatContainer.scrollHeight;

        }



        function showLoading() {

            const chatContainer = document.getElementById('chat-container');

            const loadingDiv = document.createElement('div');

            loadingDiv.className = 'loading';

            loadingDiv.id = 'loading';

            loadingDiv.textContent = 'AI is thinking...';

            chatContainer.appendChild(loadingDiv);

            chatContainer.scrollTop = chatContainer.scrollHeight;

        }



        function hideLoading() {

            const loading = document.getElementById('loading');

            if (loading) {

                loading.remove();

            }

        }



        async function sendMessage() {

            const input = document.getElementById('user-input');

            const message = input.value.trim();

            

            if (!message) return;

            

            addMessage(message, true);

            input.value = '';

            showLoading();

            

            try {

                const response = await fetch('/chat', {

                    method: 'POST',

                    headers: {

                        'Content-Type': 'application/json',

                    },

                    body: JSON.stringify({ message: message })

                });

                

                const data = await response.json();

                hideLoading();

                

                if (data.response) {

                    addMessage(data.response, false);

                } else {

                    addMessage('Sorry, I encountered an error. Please try again.', false);

                }

            } catch (error) {

                hideLoading();

                addMessage('Connection error. Please try again.', false);

            }

        }

    </script>

</body>

</html>

"""

def load_model():
    """Load the Qwen2 model and tokenizer from HF repository"""
    global model, tokenizer
    
    try:
        logger.info("Loading model and tokenizer from Hugging Face...")
        
        # Load from the deployed model repository
        model_repo = "Gaston895/Aegisecon1"
        
        logger.info(f"Loading tokenizer from {model_repo}...")
        tokenizer = AutoTokenizer.from_pretrained(
            model_repo,
            trust_remote_code=True,
            use_auth_token=False
        )
        
        logger.info(f"Loading model from {model_repo}...")
        model = AutoModelForCausalLM.from_pretrained(
            model_repo,
            torch_dtype=torch.float16,  # Changed from bfloat16 for better compatibility
            device_map="cpu",           # Force CPU for HF Spaces compatibility
            trust_remote_code=True,
            use_auth_token=False,
            low_cpu_mem_usage=True
        )
        
        logger.info("Model loaded successfully from HF repository!")
        return True
        
    except Exception as e:
        logger.error(f"Error loading model from HF: {str(e)}")
        # Try alternative loading method
        try:
            logger.info("Trying alternative loading method...")
            tokenizer = AutoTokenizer.from_pretrained(
                "Qwen/Qwen2-1.5B",  # Fallback to base model
                trust_remote_code=True
            )
            model = AutoModelForCausalLM.from_pretrained(
                "Qwen/Qwen2-1.5B",
                torch_dtype=torch.float16,
                device_map="cpu",
                trust_remote_code=True,
                low_cpu_mem_usage=True
            )
            logger.info("Fallback model loaded successfully!")
            return True
        except Exception as e2:
            logger.error(f"Fallback loading also failed: {str(e2)}")
            return False

def generate_response(prompt):
    """Generate response using the loaded model"""
    try:
        if model is None or tokenizer is None:
            return "Model is still loading, please wait a moment and try again..."
        
        # Economics-focused system prompt
        system_prompt = """You are AEGIS Economics AI, an expert economic analyst and policy advisor. 

        Provide clear, accurate, and insightful responses about economics, finance, markets, and policy.

        Focus on practical analysis and actionable insights."""
        
        full_prompt = f"{system_prompt}\n\nUser: {prompt}\nAssistant:"
        
        # Tokenize input
        inputs = tokenizer(full_prompt, return_tensors="pt", truncation=True, max_length=1024)
        
        # Generate response
        with torch.no_grad():
            outputs = model.generate(
                inputs.input_ids,
                max_new_tokens=256,  # Reduced for faster generation
                temperature=0.7,
                do_sample=True,
                pad_token_id=tokenizer.eos_token_id,
                repetition_penalty=1.1,
                no_repeat_ngram_size=3
            )
        
        # Decode response
        response = tokenizer.decode(outputs[0], skip_special_tokens=True)
        
        # Extract only the assistant's response
        if "Assistant:" in response:
            response = response.split("Assistant:")[-1].strip()
        
        return response
        
    except Exception as e:
        logger.error(f"Error generating response: {str(e)}")
        return "I apologize, but I'm having trouble processing your request right now. Please try again in a moment."

@app.route('/')
def home():
    """Serve the main chat interface"""
    return render_template_string(HTML_TEMPLATE)

@app.route('/chat', methods=['POST'])
def chat():
    """Handle chat messages"""
    try:
        data = request.get_json()
        user_message = data.get('message', '')
        
        if not user_message:
            return jsonify({'error': 'No message provided'}), 400
        
        # Generate AI response
        ai_response = generate_response(user_message)
        
        return jsonify({'response': ai_response})
        
    except Exception as e:
        logger.error(f"Error in chat endpoint: {str(e)}")
        return jsonify({'error': 'Internal server error'}), 500

@app.route('/health')
def health():
    """Health check endpoint"""
    return jsonify({
        'status': 'healthy',
        'model_loaded': model is not None,
        'tokenizer_loaded': tokenizer is not None,
        'model_info': 'Gaston895/Aegisecon1' if model is not None else 'Not loaded'
    })

@app.route('/load_model', methods=['POST'])
def load_model_endpoint():
    """Endpoint to trigger model loading"""
    try:
        success = load_model()
        return jsonify({
            'success': success,
            'model_loaded': model is not None,
            'tokenizer_loaded': tokenizer is not None
        })
    except Exception as e:
        return jsonify({'error': str(e)}), 500

if __name__ == '__main__':
    # Load model on startup
    logger.info("Starting AEGIS Economics AI...")
    
    # Try to load model, but don't fail if it doesn't work
    logger.info("Attempting to load model...")
    model_loaded = load_model()
    
    if model_loaded:
        logger.info("Model loaded successfully, starting server...")
    else:
        logger.warning("Model failed to load, starting server anyway. Model can be loaded via /load_model endpoint.")
    
    app.run(host='0.0.0.0', port=7860, debug=False)