File size: 10,283 Bytes
2ac4fcc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import os 
import gradio as gr
import pandas as pd
import torch
import logging
from transformers import pipeline, AutoModelForSequenceClassification, AutoTokenizer
import gc

# Setup logging
logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(levelname)s - %(message)s'
)
logger = logging.getLogger(__name__)

# Device configuration
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
logger.info(f"Using device: {DEVICE}")

def clear_gpu_memory():
    """Utility function to clear GPU memory"""
    if DEVICE == "cuda":
        torch.cuda.empty_cache()
    gc.collect()

class ModelManager:
    """Handles model loading and inference"""
    
    def __init__(self):
        self.device = DEVICE
        self.models = {}
        self.tokenizers = {}
    
    def load_model(self, model_name, model_type="sentiment"):
        """Load model and tokenizer"""
        try:
            if model_name not in self.models:
                if model_type == "sentiment":
                    self.tokenizers[model_name] = AutoTokenizer.from_pretrained(model_name)
                    self.models[model_name] = AutoModelForSequenceClassification.from_pretrained(
                        model_name,
                        torch_dtype=torch.float16 if self.device == "cuda" else torch.float32
                    ).to(self.device)
                else:
                    self.models[model_name] = pipeline(
                        "text-generation",
                        model=model_name,
                        device_map="auto" if self.device == "cuda" else None,
                        torch_dtype=torch.float16 if self.device == "cuda" else torch.float32
                    )
                logger.info(f"Loaded model: {model_name}")
        except Exception as e:
            logger.error(f"Error loading model {model_name}: {str(e)}")
            raise

    def unload_model(self, model_name):
        """Unload model and tokenizer"""
        try:
            if model_name in self.models:
                del self.models[model_name]
            if model_name in self.tokenizers:
                del self.tokenizers[model_name]
            clear_gpu_memory()
            logger.info(f"Unloaded model: {model_name}")
        except Exception as e:
            logger.error(f"Error unloading model {model_name}: {str(e)}")

    def get_model(self, model_name):
        """Get loaded model"""
        return self.models.get(model_name)

    def get_tokenizer(self, model_name):
        """Get loaded tokenizer"""
        return self.tokenizers.get(model_name)

class FinancialAnalyzer:
    """Main analyzer class for financial statements"""
    
    def __init__(self):
        self.model_manager = ModelManager()
        self.models = {
            "sentiment": "ProsusAI/finbert",
            "analysis": "TinyLlama/TinyLlama-1.1B-Chat-v1.0",
            "recommendation": "tiiuae/falcon-rw-1b"
        }
        
        # Load sentiment model at initialization
        try:
            self.model_manager.load_model(self.models["sentiment"], "sentiment")
        except Exception as e:
            logger.error(f"Failed to initialize sentiment model: {str(e)}")
            raise

    def read_csv(self, file_obj):
        """Read and validate CSV file"""
        try:
            if file_obj is None:
                raise ValueError("No file provided")
            
            df = pd.read_csv(file_obj)
            
            if df.empty:
                raise ValueError("Empty CSV file")
                
            return df.describe()
        except Exception as e:
            logger.error(f"Error reading CSV: {str(e)}")
            raise


    def analyze_sentiment(self, text):
        """Analyze sentiment using FinBERT"""
        try:
            model_name = self.models["sentiment"]
            model = self.model_manager.get_model(model_name)
            tokenizer = self.model_manager.get_tokenizer(model_name)

            inputs = tokenizer(
                text,
                return_tensors="pt",
                truncation=True,
                max_length=512,
                padding=True
            ).to(DEVICE)

            with torch.no_grad():
                outputs = model(**inputs)
                probabilities = torch.nn.functional.softmax(outputs.logits, dim=1)

            labels = ['negative', 'neutral', 'positive']
            scores = probabilities[0].cpu().tolist()

            results = [
                {'label': label, 'score': score}
                for label, score in zip(labels, scores)
            ]

            return [results]
        except Exception as e:
            logger.error(f"Sentiment analysis error: {str(e)}")
            return [{"label": "error", "score": 1.0}]

    def generate_analysis(self, financial_data):
        """Generate strategic analysis"""
        try:
            model_name = self.models["analysis"]
            self.model_manager.load_model(model_name, "generation")

            prompt = f"""[INST] Analyze these financial statements:
            {financial_data}
            Provide:
            1. Business Health Assessment
            2. Key Strategic Insights
            3. Market Position
            4. Growth Opportunities
            5. Risk Factors [/INST]"""

            response = self.model_manager.get_model(model_name)(
                prompt,
                max_length=1000,
                temperature=0.7,
                do_sample=True,
                num_return_sequences=1,
                truncation=True
            )

            return response[0]['generated_text']
        except Exception as e:
            logger.error(f"Analysis generation error: {str(e)}")
            return "Error in analysis generation"
        finally:
            self.model_manager.unload_model(model_name)

    def generate_recommendations(self, analysis):
        """Generate recommendations"""
        try:
            model_name = self.models["recommendation"]
            self.model_manager.load_model(model_name, "generation")

            prompt = f"""Based on this analysis:
            {analysis}
            
            Provide actionable recommendations for:
            1. Strategic Initiatives
            2. Operational Improvements
            3. Financial Management
            4. Risk Mitigation
            5. Growth Strategy"""

            response = self.model_manager.get_model(model_name)(
                prompt,
                max_length=1000,
                temperature=0.6,
                do_sample=True,
                num_return_sequences=1,
                truncation=True
            )

            return response[0]['generated_text']
        except Exception as e:
            logger.error(f"Recommendations generation error: {str(e)}")
            return "Error generating recommendations"
        finally:
            self.model_manager.unload_model(model_name)



def analyze_financial_statements(income_statement, balance_sheet):
    """Main analysis function"""
    try:
        analyzer = FinancialAnalyzer()

        # Validate inputs
        if not income_statement or not balance_sheet:
            return "Error: Please provide both income statement and balance sheet files"

        # Process financial statements
        logger.info("Processing financial statements...")
        income_summary = analyzer.read_csv(income_statement)
        balance_summary = analyzer.read_csv(balance_sheet)

        financial_data = f"""
        Income Statement Summary:
        {income_summary.to_string()}
        
        Balance Sheet Summary:
        {balance_summary.to_string()}
        """

        # Generate analysis
        logger.info("Generating analysis...")
        analysis = analyzer.generate_analysis(financial_data)

        # Analyze sentiment
        logger.info("Analyzing sentiment...")
        sentiment = analyzer.analyze_sentiment(analysis)

        # Generate recommendations
        logger.info("Generating recommendations...")
        recommendations = analyzer.generate_recommendations(analysis)

        # Format results
        return format_results(analysis, sentiment, recommendations)

    except Exception as e:
        logger.error(f"Analysis error: {str(e)}")
        return f"""Analysis Error:
        
        {str(e)}
        
        Please verify:
        1. Files are valid CSV format
        2. Files contain required financial data
        3. File size is within limits"""

def format_results(analysis, sentiment, recommendations):
    """Format analysis results"""
    try:
        if not isinstance(analysis, str) or not isinstance(recommendations, str):
            raise ValueError("Invalid input types")

        output = [
            "# Financial Analysis Report\n\n",
            "## Strategic Analysis\n\n",
            f"{analysis.strip()}\n\n",
            "## Market Sentiment\n\n"
        ]

        if isinstance(sentiment, list) and sentiment:
            for score in sentiment[0]:
                if isinstance(score, dict) and 'label' in score and 'score' in score:
                    output.append(f"- {score['label']}: {score['score']:.2%}\n")
        output.append("\n")

        output.append("## Strategic Recommendations\n\n")
        output.append(f"{recommendations.strip()}")

        return "".join(output)
    except Exception as e:
        logger.error(f"Formatting error: {str(e)}")
        return "Error formatting results"

# Create Gradio interface
iface = gr.Interface(
    fn=analyze_financial_statements,
    inputs=[
        gr.File(label="Income Statement (CSV)"),
        gr.File(label="Balance Sheet (CSV)")
    ],
    outputs=gr.Markdown(),
    title="Financial Statement Analyzer",
    description="""Upload financial statements for AI-powered analysis:
    - Strategic Analysis (TinyLlama)
    - Sentiment Analysis (FinBERT)
    - Strategic Recommendations (Falcon)
    
    Note: Please ensure files are in CSV format.""",
    flagging_mode="never"
)

if __name__ == "__main__":
    try:
        iface.queue()
        iface.launch(
            share=False,
            server_name="0.0.0.0",
            server_port=7860
        )
    except Exception as e:
        logger.error(f"Launch error: {str(e)}")
        sys.exit(1)