Update app.py

app.py

@@ -1,358 +1,57 @@
-
 import subprocess
 import sys
-import time
-import threading
-import logging
-from typing import Dict, Optional, Tuple
-import json
-import os
-from datetime import datetime, timedelta
-
-# Configure logging
-logging.basicConfig(
-    level=logging.INFO,
-    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
-)
-logger = logging.getLogger(__name__)
 
-# Dynamic installation of required packages
 def install_and_import(package):
     try:
         __import__(package)
     except ImportError:
-        logger.info(f"{package} is not installed, installing...")
+        print(f"{package} is not installed, installing...")
         subprocess.check_call([sys.executable, "-m", "pip", "install", package])
 
-# Install required packages
+# Gerekli paketleri kontrol et ve kur
 install_and_import("gradio")
 install_and_import("transformers")
 install_and_import("torch")
 
+# Şimdi import et
 import gradio as gr
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 
-# Fixed Window Rate Limiter
-class RateLimiter:
-    def __init__(self, config: Dict[str, Dict[str, int]]):
-        self.config = config
-        self.user_data = {}  # {ip: {tier: {'count': x, 'window_start': timestamp}}}
-        self.lock = threading.Lock()
-        
-    def check_and_record_request(self, identifier: str) -> Tuple[bool, Optional[str], Optional[Dict]]:
-        """Check if request is allowed and record it"""
-        with self.lock:
-            current_time = time.time()
-            
-            # Initialize user data if needed
-            if identifier not in self.user_data:
-                self.user_data[identifier] = {}
-            
-            # Check each tier
-            for tier_name, tier_config in self.config.items():
-                window_seconds = tier_config['window']
-                limit = tier_config['limit']
-                
-                # Initialize tier data if needed
-                if tier_name not in self.user_data[identifier]:
-                    self.user_data[identifier][tier_name] = {
-                        'count': 0,
-                        'window_start': current_time
-                    }
-                
-                tier_data = self.user_data[identifier][tier_name]
-                
-                # Check if current window has expired
-                window_elapsed = current_time - tier_data['window_start']
-                
-                if window_elapsed >= window_seconds:
-                    # Window expired, reset
-                    tier_data['count'] = 0
-                    tier_data['window_start'] = current_time
-                    logger.info(f"Window reset for {identifier} - {tier_name}")
-                
-                # Check if limit reached
-                if tier_data['count'] >= limit:
-                    wait_time = window_seconds - window_elapsed
-                    
-                    error_msg = (
-                        f"Rate limit aşıldı ({tier_name}): "
-                        f"{tier_data['count']}/{limit} kullanıldı. "
-                        f"Bekleme süresi: {wait_time:.0f} saniye"
-                    )
-                    
-                    stats = self._get_all_stats(identifier, current_time)
-                    return False, error_msg, stats
-            
-            # Request allowed - increment all counters
-            for tier_name in self.config:
-                self.user_data[identifier][tier_name]['count'] += 1
-            
-            stats = self._get_all_stats(identifier, current_time)
-            logger.info(f"Request allowed for {identifier}. Stats: {stats}")
-            
-            return True, None, stats
-    
-    def _get_all_stats(self, identifier: str, current_time: float) -> Dict:
-        """Get usage statistics for all tiers"""
-        stats = {}
-        
-        for tier_name, tier_config in self.config.items():
-            window_seconds = tier_config['window']
-            limit = tier_config['limit']
-            
-            if identifier in self.user_data and tier_name in self.user_data[identifier]:
-                tier_data = self.user_data[identifier][tier_name]
-                window_elapsed = current_time - tier_data['window_start']
-                
-                # If window expired, show as reset
-                if window_elapsed >= window_seconds:
-                    used = 0
-                    reset_in = 0
-                else:
-                    used = tier_data['count']
-                    reset_in = window_seconds - window_elapsed
-            else:
-                used = 0
-                reset_in = 0
-            
-            stats[tier_name] = {
-                'used': used,
-                'limit': limit,
-                'remaining': max(0, limit - used),
-                'reset_in': reset_in,
-                'window_text': self._format_window(window_seconds)
-            }
-        
-        return stats
-    
-    def _format_window(self, seconds: int) -> str:
-        """Format time window in Turkish"""
-        if seconds < 60:
-            return f"{seconds} saniye"
-        elif seconds < 3600:
-            return f"{seconds // 60} dakika"
-        elif seconds < 86400:
-            return f"{seconds // 3600} saat"
-        else:
-            return f"{seconds // 86400} gün"
-
-# Initialize model
-logger.info("Model yükleniyor...")
+# Model ve tokenizer yükleme 'yeni
 model_name = "Bertug1911/BrtGPT-124m-Base"
-
-try:
-    tokenizer = AutoTokenizer.from_pretrained(model_name)
-    model = AutoModelForCausalLM.from_pretrained(model_name)
-    model.eval()
-    
-    if torch.cuda.is_available():
-        model.to("cuda")
-        logger.info("Model CUDA'da yüklendi")
-    else:
-        logger.info("Model CPU'da yüklendi")
-except Exception as e:
-    logger.error(f"Model yükleme hatası: {e}")
-    raise
-
-# Rate limiting configuration
-RATE_LIMIT_CONFIG = {
-    'dakika': {'window': 60, 'limit': 5},       # 5 istek / dakika
-    'saat': {'window': 3600, 'limit': 50},      # 50 istek / saat
-    'gün': {'window': 86400, 'limit': 500},     # 500 istek / gün
-}
-
-# Initialize rate limiter
-rate_limiter = RateLimiter(RATE_LIMIT_CONFIG)
-
-# Generation function
-def generate_text_with_rate_limit(
-    prompt: str, 
-    temperature: float, 
-    top_k: int, 
-    max_new_tokens: int, 
-    request: gr.Request
-) -> str:
-    """Generate text with rate limiting"""
-    
-    # Get client IP
-    ip = request.client.host if request and request.client else "unknown"
-    
-    # Check rate limit
-    is_allowed, error_msg, stats = rate_limiter.check_and_record_request(ip)
-    
-    if not is_allowed:
-        error_output = f"❌ **{error_msg}**\n\n"
-        error_output += "📊 **Kredi Durumu:**\n\n"
-        
-        for tier, info in stats.items():
-            percentage = (info['remaining'] / info['limit']) * 100 if info['limit'] > 0 else 0
-            
-            if percentage > 50:
-                bar = "🟩" * 5
-            elif percentage > 20:
-                bar = "🟨" * 3 + "⬜" * 2
-            else:
-                bar = "🟥" * 1 + "⬜" * 4
-            
-            error_output += f"**{tier.capitalize()}** [{bar}]\n"
-            error_output += f"├─ Kullanılan: {info['used']}/{info['limit']}\n"
-            error_output += f"├─ Kalan: {info['remaining']}\n"
-            error_output += f"└─ Yenileme: {info['reset_in']:.0f} saniye\n\n"
-        
-        return error_output
-    
-    try:
-        # Input validation
-        if not prompt or len(prompt.strip()) == 0:
-            return "⚠️ Lütfen bir prompt girin."
-        
-        if len(prompt) > 1000:
-            return "⚠️ Prompt çok uzun. Maksimum 1000 karakter."
-        
-        # Generate text
-        logger.info(f"Metin oluşturuluyor: temp={temperature}, top_k={top_k}, max_tokens={max_new_tokens}")
-        
-        inputs = tokenizer(prompt, return_tensors="pt", truncation=True, max_length=512)
-        if torch.cuda.is_available():
-            inputs = inputs.to("cuda")
-        
-        with torch.no_grad():
-            output = model.generate(
-                **inputs,
-                max_new_tokens=int(max_new_tokens),
-                temperature=float(temperature),
-                top_k=int(top_k),
-                do_sample=True,
-                pad_token_id=tokenizer.eos_token_id
-            )
-        
-        generated_text = tokenizer.decode(output[0], skip_special_tokens=False)
-        generated_text = generated_text.replace("Ġ", " ")
-        
-        # Add credit status
-        footer = "\n\n" + "─" * 50 + "\n"
-        footer += "💳 **Kredi Durumu:**\n\n"
-        
-        for tier, info in stats.items():
-            percentage = (info['remaining'] / info['limit']) * 100 if info['limit'] > 0 else 0
-            
-            if percentage > 50:
-                emoji = "✅"
-            elif percentage > 20:
-                emoji = "⚠️"
-            else:
-                emoji = "🔴"
-            
-            footer += f"{emoji} **{tier.capitalize()}**: "
-            footer += f"{info['used']}/{info['limit']} kullanıldı "
-            footer += f"({info['remaining']} kaldı)"
-            
-            if info['reset_in'] > 0:
-                minutes = int(info['reset_in'] // 60)
-                seconds = int(info['reset_in'] % 60)
-                if minutes > 0:
-                    footer += f" - Yenileme: {minutes}dk {seconds}sn"
-                else:
-                    footer += f" - Yenileme: {seconds} saniye"
-            footer += "\n"
-        
-        return generated_text + footer
-        
-    except Exception as e:
-        logger.error(f"Metin oluşturma hatası: {e}", exc_info=True)
-        return f"❌ Hata: {str(e)}"
-
-# Gradio interface
-with gr.Blocks(title="BrtGPT-124m-Base", theme=gr.themes.Soft()) as app:
-    gr.Markdown("""
-    # 🤖 BrtGPT-124m-Base Metin Üreteci
-    
-    ### 📋 Kredi Sistemi (Fixed Window)
-    - **Dakika**: 5 istek / 60 saniye
-    - **Saat**: 50 istek / 3600 saniye  
-    - **Gün**: 500 istek / 86400 saniye
-    
-    ⚠️ **Not**: Her kategori kendi pencere süresine göre sıfırlanır. 
-    İlk kullanımdan itibaren süre başlar.
-    """)
-    
-    with gr.Row():
-        with gr.Column(scale=2):
-            prompt_input = gr.Textbox(
-                lines=5,
-                placeholder="Metin üretmek için prompt'unuzu buraya yazın...",
-                label="📝 Prompt",
-                max_lines=10
-            )
-            
-            with gr.Row():
-                temperature_slider = gr.Slider(
-                    minimum=0.01,
-                    maximum=1.0,
-                    value=0.5,
-                    step=0.01,
-                    label="🌡️ Temperature (Yaratıcılık)"
-                )
-                top_k_slider = gr.Slider(
-                    minimum=1,
-                    maximum=50,
-                    value=10,
-                    step=1,
-                    label="🎯 Top-K (Çeşitlilik)"
-                )
-                max_tokens_slider = gr.Slider(
-                    minimum=1,
-                    maximum=100,
-                    value=30,
-                    step=1,
-                    label="📏 Maksimum Token"
-                )
-            
-            generate_button = gr.Button("🚀 Metin Oluştur", variant="primary", size="lg")
-        
-        with gr.Column(scale=3):
-            output_text = gr.Textbox(
-                label="📄 Oluşturulan Metin",
-                lines=15,
-                max_lines=25,
-                interactive=False,
-                show_copy_button=True
-            )
-    
-    gr.Markdown("""
-    ---
-    ### ℹ️ Sistem Bilgisi
-    - Toplam 3 paralel instance mevcut
-    - Şu an **Instance 1**'desiniz
-    - Yoğunluk durumunda diğer instance'ları deneyin
-    
-    📧 **İletişim**: bertug2099@gmail.com | bertugscpmail@gmail.com
-    """)
-    
-    # Event handler
-    generate_button.click(
-        fn=generate_text_with_rate_limit,
-        inputs=[prompt_input, temperature_slider, top_k_slider, max_tokens_slider],
-        outputs=output_text
-    )
-    
-    # Examples
-    gr.Examples(
-        examples=[
-            ["Teknolojinin gelecekte hayatımızı nasıl değiştireceğini", 0.7, 15, 50],
-            ["Gizemli bir ormanın derinliklerinde", 0.5, 10, 40],
-            ["Uzay yolculuğu ve insanlığın geleceği", 0.6, 20, 60],
-        ],
-        inputs=[prompt_input, temperature_slider, top_k_slider, max_tokens_slider],
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)
+
+def generate_text(prompt, temperature, top_k, max_new_tokens):
+    inputs = tokenizer(prompt, return_tensors="pt")
+    output = model.generate(
+        **inputs,
+        max_new_tokens=int(max_new_tokens),
+        temperature=float(temperature),
+        top_k=int(top_k),
+        do_sample=True,
     )
+    generated_text = tokenizer.decode(output[0], skip_special_tokens=False)
+    generated_text = generated_text.replace(" ", "").replace("Ġ", " ")
+    return generated_text
+
+arayuz = gr.Interface(
+    fn=generate_text,
+    inputs=[
+        gr.Textbox(lines=3, placeholder="Your prompt..."),
+        gr.Slider(minimum=0.01, maximum=1.0, value=0.5, step=0.01, label="Temperature"),
+        gr.Slider(minimum=1, maximum=50, value=10, step=1, label="Top-K"),
+        gr.Slider(minimum=1, maximum=50, value=15, step=1, label="Max New Tokens"),
+    ],
+    outputs="text",
+    title="BrtGPT-124m-Base",
+    description="""
+    If the system/application slows down as the number of users increases, we will update the number and type of GPUs. 
+    Also, until we update, there are 3 applications that do the same thing but are copied to distribute the workload. 
+    "YOU ARE CURRENTLY ON APPLICATION 1" 
+    If this application slows down too much, you can use the 2nd or 3rd application, if you think all of them are slowing down right now, contact me: 
+    "bertug2099@gmail.com or bertugscpmail@gmail.com"""
+)
 
-if __name__ == "__main__":
-    logger.info("Uygulama başlatılıyor...")
-    app.launch(
-        server_name="0.0.0.0",
-        server_port=7860,
-        share=False,
-        show_error=True
-    )
+arayuz.launch()