Spaces:
Sleeping
Sleeping
File size: 22,522 Bytes
ea4b18b b2501de 0ad8f2f ab0bea5 0ad8f2f 6631d2e 74f7ba2 0ad8f2f 6631d2e ea4b18b 5bc4f16 de70888 0ad8f2f 5bc4f16 0ad8f2f 6631d2e 0ad8f2f 6631d2e 0ad8f2f 6631d2e 5bc4f16 6631d2e 5bc4f16 6631d2e 5bc4f16 6ef47af 6631d2e 88b54ed 6631d2e cd3edfb 6631d2e 88b54ed 6631d2e 0ad8f2f 6631d2e 88b54ed 6631d2e 88b54ed cd3edfb 6631d2e 5b6a3e1 c518467 cd3edfb 6631d2e cd3edfb c518467 db53aac cd3edfb db53aac cd3edfb c518467 5bc4f16 80ca468 0ad8f2f cd3edfb 0ad8f2f 80ca468 0ad8f2f 80ca468 0ad8f2f 80ca468 0ad8f2f 80ca468 0ad8f2f cd3edfb 0ad8f2f cd3edfb 0ad8f2f cd3edfb 0ad8f2f cd3edfb 5bc4f16 de70888 0ad8f2f 5bc4f16 de70888 0ad8f2f de70888 0ad8f2f 6631d2e de70888 5bc4f16 de70888 74f7ba2 0ad8f2f 74f7ba2 c6b42a6 9bdd84e 0ad8f2f 6ef47af 0ad8f2f ab0bea5 5bc4f16 ab0bea5 5bc4f16 c6b42a6 0ad8f2f 6631d2e de70888 0ad8f2f 6631d2e c6b42a6 0ad8f2f c6b42a6 5bc4f16 0ad8f2f 6631d2e 5bc4f16 9bdd84e ab0bea5 c6b42a6 74f7ba2 9bdd84e 6631d2e 9bdd84e 5462ac3 6631d2e ea4b18b 74f7ba2 ea4b18b c6b42a6 ea4b18b ab0bea5 0ad8f2f ea4b18b 5462ac3 ea4b18b 5462ac3 cd3edfb |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 |
import gradio as gr
import pandas as pd
import json
from transformers import (
AutoTokenizer,
AutoModelForCausalLM,
AutoModelForSequenceClassification,
TrainingArguments,
Trainer
)
import torch
import numpy as np
from torch.utils.data import Dataset, DataLoader
import re
class FinancialDataset(Dataset):
def __init__(self, texts, labels, tokenizer, max_length=512):
self.texts = texts
self.labels = labels
self.tokenizer = tokenizer
self.max_length = max_length
def __len__(self):
return len(self.texts)
def __getitem__(self, idx):
text = str(self.texts[idx])
inputs = self.tokenizer(
text,
truncation=True,
padding='max_length',
max_length=self.max_length,
return_tensors='pt'
)
return {
'input_ids': inputs['input_ids'].squeeze(),
'attention_mask': inputs['attention_mask'].squeeze(),
'labels': torch.tensor(self.labels[idx], dtype=torch.long)
}
class FinancialAnalyzer:
def __init__(self):
print("Initializing Analyzer...")
self.last_metrics = {}
self.initialize_models()
print("Initialization complete!")
def initialize_models(self):
"""Initialize both TinyLlama and FinBERT models"""
try:
# Initialize TinyLlama
self.llama_tokenizer = AutoTokenizer.from_pretrained("TinyLlama/TinyLlama-1.1B-Chat-v1.0")
self.llama_model = AutoModelForCausalLM.from_pretrained("TinyLlama/TinyLlama-1.1B-Chat-v1.0")
self.llama_model.eval()
# Initialize FinBERT
self.finbert_tokenizer = AutoTokenizer.from_pretrained("ProsusAI/finbert")
self.finbert_model = AutoModelForSequenceClassification.from_pretrained("ProsusAI/finbert")
self.finbert_model.eval()
print("Models loaded successfully!")
except Exception as e:
print(f"Error initializing models: {str(e)}")
raise
def clean_number(self, value):
"""Clean and convert numerical values"""
try:
if isinstance(value, str):
value = value.replace('$', '').replace(',', '').strip()
if '(' in value and ')' in value:
value = '-' + value.replace('(', '').replace(')', '')
return float(value or 0)
except:
return 0.0
def is_valid_markdown(self, file_path):
"""Check if a file is a valid Markdown file"""
try:
with open(file_path, 'r') as f:
content = f.read()
return any(line.startswith('#') or '|' in line for line in content.split('\n'))
except:
return False
def parse_financial_data(self, content):
"""Parse markdown content into structured data"""
try:
data = {}
current_section = ""
current_table = []
headers = None
for line in content.split('\n'):
if line.startswith('#'):
if current_table and headers:
data[current_section] = self.process_table(headers, current_table)
current_section = line.strip('# ')
current_table = []
headers = None
elif '|' in line:
if '-|-' not in line:
row = [cell.strip() for cell in line.split('|')[1:-1]]
if not headers:
headers = row
else:
current_table.append(row)
if current_table and headers:
data[current_section] = self.process_table(headers, current_table)
return data
except Exception as e:
print(f"Error parsing financial data: {str(e)}")
return {}
def process_table(self, headers, rows):
"""Process table data into structured format"""
try:
processed_data = {}
for row in rows:
if len(row) == len(headers):
item_name = row[0].strip('*').strip()
processed_data[item_name] = {}
for i, value in enumerate(row[1:], 1):
processed_data[item_name][headers[i]] = self.clean_number(value)
return processed_data
except Exception as e:
print(f"Error processing table: {str(e)}")
return {}
def get_nested_value(self, data, section, key, year):
"""Safely get nested dictionary value"""
try:
return data.get(section, {}).get(key, {}).get(str(year), 0)
except:
return 0
def extract_metrics(self, income_data, balance_data):
"""Extract and calculate key financial metrics"""
try:
metrics = {
"Revenue": {
"2025": self.get_nested_value(income_data, "Revenue", "Total Net Revenue", "2025"),
"2024": self.get_nested_value(income_data, "Revenue", "Total Net Revenue", "2024"),
"2021": self.get_nested_value(income_data, "Revenue", "Total Net Revenue", "2021")
},
"Profitability": {
"Gross_Profit_2025": self.get_nested_value(income_data, "Cost and Gross Profit", "Gross Profit", "2025"),
"EBIT_2025": self.get_nested_value(income_data, "Profit Summary", "EBIT", "2025"),
"Net_Earnings_2025": self.get_nested_value(income_data, "Profit Summary", "Net Earnings", "2025"),
"Operating_Expenses_2025": self.get_nested_value(income_data, "Operating Expenses", "Total Operating Expenses", "2025")
},
"Balance_Sheet": {
"Total_Assets_2025": self.get_nested_value(balance_data, "Key Totals", "Total_Assets", "2025"),
"Current_Assets_2025": self.get_nested_value(balance_data, "Key Totals", "Total_Current_Assets", "2025"),
"Total_Liabilities_2025": self.get_nested_value(balance_data, "Key Totals", "Total_Liabilities", "2025"),
"Current_Liabilities_2025": self.get_nested_value(balance_data, "Key Totals", "Total_Current_Liabilities", "2025"),
"Equity_2025": self.get_nested_value(balance_data, "Key Totals", "Total_Shareholders_Equity", "2025"),
"Inventory_2025": self.get_nested_value(balance_data, "Balance Sheet Data 2021-2025", "Inventory", "2025"),
"Accounts_Receivable_2025": self.get_nested_value(balance_data, "Balance Sheet Data 2021-2025", "Accounts_Receivable", "2025"),
"Long_Term_Debt_2025": self.get_nested_value(balance_data, "Balance Sheet Data 2021-2025", "Long_Term_Debt", "2025")
},
"Cash_Flow": {
"Depreciation_2025": self.get_nested_value(income_data, "Operating Expenses", "Depreciation & Amortization", "2025"),
"Interest_Expense_2025": self.get_nested_value(income_data, "Profit Summary", "Interest Expense", "2025")
}
}
revenue_2025 = metrics["Revenue"]["2025"]
if revenue_2025 != 0:
metrics["Ratios"] = {
"Gross_Margin": (metrics["Profitability"]["Gross_Profit_2025"] / revenue_2025) * 100,
"Operating_Margin": (metrics["Profitability"]["EBIT_2025"] / revenue_2025) * 100,
"Net_Margin": (metrics["Profitability"]["Net_Earnings_2025"] / revenue_2025) * 100,
"Current_Ratio": metrics["Balance_Sheet"]["Current_Assets_2025"] / metrics["Balance_Sheet"]["Current_Liabilities_2025"] if metrics["Balance_Sheet"]["Current_Liabilities_2025"] != 0 else 0,
"Quick_Ratio": (metrics["Balance_Sheet"]["Current_Assets_2025"] - metrics["Balance_Sheet"]["Inventory_2025"]) / metrics["Balance_Sheet"]["Current_Liabilities_2025"] if metrics["Balance_Sheet"]["Current_Liabilities_2025"] != 0 else 0,
"Asset_Turnover": revenue_2025 / metrics["Balance_Sheet"]["Total_Assets_2025"] if metrics["Balance_Sheet"]["Total_Assets_2025"] != 0 else 0,
"Receivables_Turnover": revenue_2025 / metrics["Balance_Sheet"]["Accounts_Receivable_2025"] if metrics["Balance_Sheet"]["Accounts_Receivable_2025"] != 0 else 0,
"Debt_to_Equity": metrics["Balance_Sheet"]["Total_Liabilities_2025"] / metrics["Balance_Sheet"]["Equity_2025"] if metrics["Balance_Sheet"]["Equity_2025"] != 0 else 0,
"Interest_Coverage": metrics["Profitability"]["EBIT_2025"] / metrics["Cash_Flow"]["Interest_Expense_2025"] if metrics["Cash_Flow"]["Interest_Expense_2025"] != 0 else 0,
"Revenue_Growth": ((metrics["Revenue"]["2025"] / metrics["Revenue"]["2024"]) - 1) * 100 if metrics["Revenue"]["2024"] != 0 else 0,
"5Year_Revenue_CAGR": ((metrics["Revenue"]["2025"] / metrics["Revenue"]["2021"]) ** (1/4) - 1) * 100 if metrics["Revenue"]["2021"] != 0 else 0
}
return metrics
except Exception as e:
print(f"Error extracting metrics: {str(e)}")
return {}
def convert_to_serializable(obj):
"""Convert numpy values to Python native types"""
if isinstance(obj, np.float32):
return float(obj)
elif isinstance(obj, np.ndarray):
return obj.tolist()
elif isinstance(obj, dict):
return {key: convert_to_serializable(value) for key, value in obj.items()}
elif isinstance(obj, list):
return [convert_to_serializable(item) for item in obj]
return obj
def get_sentiment_analysis(self, metrics):
"""Get financial sentiment analysis using FinBERT"""
try:
financial_text = f"""
Revenue growth: {metrics['Ratios'].get('Revenue_Growth', 0):.2f}%
Profit margin: {metrics['Ratios'].get('Net_Margin', 0):.2f}%
Debt to equity: {metrics['Ratios'].get('Debt_to_Equity', 0):.2f}
Interest coverage: {metrics['Ratios'].get('Interest_Coverage', 0):.2f}
Current ratio: {metrics['Ratios'].get('Current_Ratio', 0):.2f}
"""
inputs = self.finbert_tokenizer(financial_text, return_tensors="pt", padding=True, truncation=True)
outputs = self.finbert_model(**inputs)
probabilities = torch.nn.functional.softmax(outputs.logits, dim=-1)
sentiment_scores = probabilities.detach().numpy()[0]
sentiments = ['negative', 'neutral', 'positive']
sentiment_dict = dict(zip(sentiments, [float(score) for score in sentiment_scores]))
return sentiment_dict
except Exception as e:
print(f"Error in sentiment analysis: {str(e)}")
return {}
def analyze_financials(self, balance_sheet_file, income_stmt_file):
"""Main analysis function"""
try:
# Validate input files
if not (self.is_valid_markdown(balance_sheet_file) and self.is_valid_markdown(income_stmt_file)):
return "Error: One or both files are invalid or not in Markdown format."
# Read files
with open(balance_sheet_file, 'r') as f:
balance_sheet = f.read()
with open(income_stmt_file, 'r') as f:
income_stmt = f.read()
# Process financial data
income_data = self.parse_financial_data(income_stmt)
balance_data = self.parse_financial_data(balance_sheet)
metrics = self.extract_metrics(income_data, balance_data)
# Get sentiment analysis
sentiment_dict = self.get_sentiment_analysis(metrics)
# Generate and get analysis
prompt = self.generate_prompt(metrics, sentiment_dict)
analysis = self.generate_analysis(prompt)
# Convert all numpy values to Python native types
metrics = convert_to_serializable(metrics)
sentiment_dict = convert_to_serializable(sentiment_dict)
# Prepare final results
results = {
"Financial Analysis": {
"Key Metrics": metrics,
"Market Sentiment": sentiment_dict,
"AI Insights": analysis,
"Analysis Period": "2021-2025",
"Note": "All monetary values in millions ($M)"
}
}
return json.dumps(results, indent=2)
except Exception as e:
return f"Error in analysis: {str(e)}\n\nDetails: {type(e).__name__}"
def generate_prompt(self, metrics, sentiment_dict):
"""Create enhanced analysis prompt with sentiment"""
try:
return f"""[INST] As a financial analyst, provide a comprehensive analysis of this company's performance.
Financial Metrics (2025):
------------------------
1. Revenue & Growth:
- Revenue: ${metrics['Revenue']['2025']:,.1f}M
- Growth Rate: {metrics['Ratios'].get('Revenue_Growth', 0):,.1f}%
- 5-Year CAGR: {metrics['Ratios'].get('5Year_Revenue_CAGR', 0):,.1f}%
2. Profitability:
- Gross Profit: ${metrics['Profitability']['Gross_Profit_2025']:,.1f}M
- EBIT: ${metrics['Profitability']['EBIT_2025']:,.1f}M
- Net Earnings: ${metrics['Profitability']['Net_Earnings_2025']:,.1f}M
- Margins:
* Gross: {metrics['Ratios'].get('Gross_Margin', 0):,.1f}%
* Operating: {metrics['Ratios'].get('Operating_Margin', 0):,.1f}%
* Net: {metrics['Ratios'].get('Net_Margin', 0):,.1f}%
3. Financial Position:
- Assets: ${metrics['Balance_Sheet']['Total_Assets_2025']:,.1f}M
- Liabilities: ${metrics['Balance_Sheet']['Total_Liabilities_2025']:,.1f}M
- Equity: ${metrics['Balance_Sheet']['Equity_2025']:,.1f}M
4. Key Ratios:
- Liquidity: Current Ratio {metrics['Ratios'].get('Current_Ratio', 0):,.2f}x
- Efficiency: Asset Turnover {metrics['Ratios'].get('Asset_Turnover', 0):,.2f}x
- Solvency: Debt/Equity {metrics['Ratios'].get('Debt_to_Equity', 0):,.2f}x
- Coverage: Interest Coverage {metrics['Ratios'].get('Interest_Coverage', 0):,.2f}x
Market Sentiment Indicators:
---------------------------
- Positive: {sentiment_dict.get('positive', 0):,.2f}
- Neutral: {sentiment_dict.get('neutral', 0):,.2f}
- Negative: {sentiment_dict.get('negative', 0):,.2f}
Provide:
1. Overall financial health assessment
2. Key strengths and concerns
3. Operational efficiency analysis
4. Recommendations for improvement
[/INST]"""
except Exception as e:
print(f"Error generating prompt: {str(e)}")
return ""
def generate_analysis(self, prompt):
"""Generate analysis using TinyLlama"""
try:
# Format the prompt in TinyLlama's expected format
formatted_prompt = f"<human>: {prompt}\n<assistant>: Let me analyze these financial metrics in detail."
inputs = self.llama_tokenizer(
formatted_prompt,
return_tensors="pt",
truncation=True,
max_length=2048,
padding=True
)
# Generate with adjusted parameters
outputs = self.llama_model.generate(
inputs["input_ids"],
max_new_tokens=1024,
min_new_tokens=200, # Ensure minimum length
temperature=0.8, # Slightly increased creativity
top_p=0.92, # Slightly increased diversity
do_sample=True,
repetition_penalty=1.2,
length_penalty=1.5, # Encourage longer generations
num_return_sequences=1,
pad_token_id=self.llama_tokenizer.eos_token_id,
eos_token_id=self.llama_tokenizer.eos_token_id,
early_stopping=True
)
# Decode and clean up the response
analysis = self.llama_tokenizer.decode(outputs[0], skip_special_tokens=False)
# Extract only the assistant's response
if "<assistant>:" in analysis:
analysis = analysis.split("<assistant>:")[-1].strip()
# Clean up any remaining tags
analysis = analysis.replace("<human>:", "").replace("<assistant>:", "").strip()
# Validate output length and content
if len(analysis.split()) < 100:
# Fallback analysis if model generation is too short
analysis = self.generate_fallback_analysis(self.last_metrics)
return analysis
except Exception as e:
print(f"Detailed error in generate_analysis: {str(e)}")
return self.generate_fallback_analysis(self.last_metrics)
def generate_fallback_analysis(self, metrics):
"""Generate a basic analysis when the model fails"""
try:
revenue_growth = metrics['Ratios'].get('Revenue_Growth', 0)
net_margin = metrics['Ratios'].get('Net_Margin', 0)
current_ratio = metrics['Ratios'].get('Current_Ratio', 0)
debt_to_equity = metrics['Ratios'].get('Debt_to_Equity', 0)
analysis = f"""
Financial Analysis Summary:
1. Revenue and Growth:
The company shows a revenue growth of {revenue_growth:.1f}%, indicating {
'strong' if revenue_growth > 5 else 'moderate' if revenue_growth > 0 else 'weak'} growth performance.
2. Profitability:
With a net margin of {net_margin:.1f}%, the company demonstrates {
'strong' if net_margin > 10 else 'moderate' if net_margin > 5 else 'concerning'} profitability levels.
3. Liquidity Position:
The current ratio of {current_ratio:.2f}x suggests {
'very strong' if current_ratio > 2 else 'adequate' if current_ratio > 1 else 'concerning'} liquidity position.
4. Financial Leverage:
With a debt-to-equity ratio of {debt_to_equity:.2f}, the company maintains {
'conservative' if debt_to_equity < 0.5 else 'moderate' if debt_to_equity < 1 else 'aggressive'} leverage.
Key Recommendations:
1. {'Consider debt reduction' if debt_to_equity > 0.5 else 'Maintain current debt levels'}
2. {'Focus on improving profit margins' if net_margin < 5 else 'Maintain profit efficiency'}
3. {'Implement growth strategies' if revenue_growth < 2 else 'Sustain growth momentum'}
This analysis is based on key financial metrics and standard industry benchmarks.
"""
return analysis
except Exception as e:
return f"Error generating fallback analysis: {str(e)}"
def fine_tune_models(self, train_texts, train_labels, epochs=3):
"""Fine-tune the models with custom data"""
try:
# Prepare dataset
train_dataset = FinancialDataset(train_texts, train_labels, self.llama_tokenizer)
# Training arguments
training_args = TrainingArguments(
output_dir="./financial_model_tuned",
num_train_epochs=epochs,
per_device_train_batch_size=4,
logging_dir="./logs",
logging_steps=10,
save_steps=50,
eval_steps=50,
evaluation_strategy="steps",
learning_rate=2e-5,
weight_decay=0.01,
warmup_steps=500,
)
# Initialize trainer
trainer = Trainer(
model=self.llama_model,
args=training_args,
train_dataset=train_dataset,
)
# Fine-tune the model
trainer.train()
# Save the fine-tuned model
self.llama_model.save_pretrained("./financial_model_tuned")
self.llama_tokenizer.save_pretrained("./financial_model_tuned")
print("Fine-tuning completed successfully!")
except Exception as e:
print(f"Error in fine-tuning: {str(e)}")
def analyze_financials(self, balance_sheet_file, income_stmt_file):
"""Main analysis function"""
try:
# Validate input files
if not (self.is_valid_markdown(balance_sheet_file) and self.is_valid_markdown(income_stmt_file)):
return "Error: One or both files are invalid or not in Markdown format."
# Read files
with open(balance_sheet_file, 'r') as f:
balance_sheet = f.read()
with open(income_stmt_file, 'r') as f:
income_stmt = f.read()
# Process financial data
income_data = self.parse_financial_data(income_stmt)
balance_data = self.parse_financial_data(balance_sheet)
metrics = self.extract_metrics(income_data, balance_data)
self.last_metrics = metrics
# Get sentiment analysis
sentiment_dict = self.get_sentiment_analysis(metrics)
# Generate and get analysis
prompt = self.generate_prompt(metrics, sentiment_dict)
analysis = self.generate_analysis(prompt)
# Prepare final results
results = {
"Financial Analysis": {
"Key Metrics": metrics,
"Market Sentiment": sentiment_dict,
"AI Insights": analysis,
"Analysis Period": "2021-2025",
"Note": "All monetary values in millions ($M)"
}
}
return json.dumps(results, indent=2)
except Exception as e:
return f"Error in analysis: {str(e)}\n\nDetails: {type(e).__name__}"
def create_interface():
analyzer = FinancialAnalyzer()
iface = gr.Interface(
fn=analyzer.analyze_financials,
inputs=[
gr.File(label="Balance Sheet (Markdown)", type="filepath"),
gr.File(label="Income Statement (Markdown)", type="filepath")
],
outputs=gr.Textbox(label="Analysis Results", lines=25),
title="AI Financial Statement Analyzer",
description="""Upload financial statements in Markdown format for AI-powered analysis.
The analysis combines LLM-based insights with sentiment analysis."""
)
return iface
if __name__ == "__main__":
iface = create_interface()
iface.launch() |