Update README.md

README.md

@@ -9,138 +9,88 @@ app_file: app.py
 pinned: false
 ---
 
-# AI-Powered Bilingual Storyteller & Illustrator
-
-## Overview
-
-This application generates high-quality stories in both English and Arabic with emotional analysis and optional illustrations. It uses a robust template-based approach combined with AI models to ensure culturally appropriate, engaging, and safe content generation.
-
-## Key Features
-
-### 1. Reliable Bilingual Story Generation
-- **English Stories**: High-quality narrative generation with emotional analysis
-- **Arabic Stories**: Template-based system with culturally appropriate content
-- **Automatic Language Detection**: Seamlessly handles input in either language
-
-### 2. Multiple Creation Modes
-- **Basic Story Mode**: Generate stories from simple prompts
-- **Template Story Mode**: Guided creation using structured templates
-- **Visual Story Mode**: Create stories with illustrated scenes
-
-### 3. Advanced Visualization
-- Generate scene sequences from stories (1-5 scenes)
-- Multiple artistic styles: realistic, anime, fantasy
-- Automatic prompt enhancement for better image quality
-
-### 4. Content Safety System
-- Multi-layered content filtering to prevent inappropriate material
-- Language consistency verification
-- Repetition detection to maintain story quality
-- Graceful fallbacks to ensure reliable output
-
-## Technical Implementation
-
-### Story Generation Architecture
-
-The system uses a hybrid approach to story generation:
-
-1. **English Generation**:
-   - Uses EleutherAI/gpt-neo-1.3B with optimization for storytelling
-   - Enhanced with template options for consistency
-
-2. **Arabic Generation**:
-   - Template-based system with curated high-quality narratives
-   - Dynamic template selection based on prompt analysis
-   - Parameter extraction to customize stories
-   - Multiple fallback mechanisms to ensure appropriate content
-
-3. **Emotion Analysis**:
-   - English: distilbert-based sentiment analysis
-   - Arabic: CAMeL-Lab/bert-base-arabic-sentiment when available
-   - Cross-lingual sentiment analysis for comprehensive coverage
-
-4. **Translation Capabilities**:
-   - Arabic-to-English: Helsinki-NLP/opus-mt-ar-en
-   - English-to-Arabic: Helsinki-NLP/opus-mt-en-ar (when available)
-   - Used for cross-lingual operations and image generation
+# AI-Powered Bilingual Storyteller & Illustrator - Technical Summary
+
+## Core Functionality
+- Generates stories in English and Arabic with emotional analysis and optional illustrations
+- Uses template-based approach with AI models to ensure quality and safety
+
+## Technical Architecture
+
+### Story Generation
+
+#### NLP Pipelines
+- **English Text Generation Pipeline**:
+  ```python
+  pipeline("text-generation", model="EleutherAI/gpt-neo-1.3B", device="cpu")
+  ```
+
+- **Arabic Generation**:
+  ```python
+  # Uses MT5 instead of standard pipeline
+  AutoTokenizer.from_pretrained("google/mt5-small")
+  AutoModelForSeq2SeqLM.from_pretrained("google/mt5-small")
+  ```
+
+- **Sentiment Analysis Pipelines**:
+  ```python
+  # English
+  pipeline("sentiment-analysis", model="distilbert-base-uncased-finetuned-sst-2-english", device="cpu")
+  
+  # Arabic
+  pipeline("sentiment-analysis", model="CAMeL-Lab/bert-base-arabic-sentiment", device="cpu")
+  ```
+
+- **Translation Pipelines**:
+  ```python
+  # Arabic to English
+  pipeline("translation", model="Helsinki-NLP/opus-mt-ar-en", device="cpu")
+  
+  # English to Arabic
+  pipeline("translation", model="Helsinki-NLP/opus-mt-en-ar", device="cpu")
+  ```
 
 ### Visual Generation
-
-The application uses Stable Diffusion (runwayml/stable-diffusion-v1-5) for image generation with:
-
-- Efficient GPU resource management
+- **Image Generation Pipeline**:
+  ```python
+  pipe = StableDiffusionPipeline.from_pretrained(
+      "runwayml/stable-diffusion-v1-5",
+      torch_dtype=torch.float16
+  )
+  ```
+- Efficient GPU resource management via @spaces.GPU decorator
 - Scene extraction from story content
-- Style-specific prompt enhancement
-- Comprehensive error handling
-
-## Usage Instructions
-
-### Basic Story Generation
-1. Enter a prompt in English or Arabic
-2. Select your desired output language
-3. Click "Generate Story"
-4. Review your story with emotional analysis
-
-### Template Story Creation
-1. Choose a template type (Adventure, Friendship, Fantasy)
-2. Fill in the template parameters or use defaults
-3. Select output language
-4. Generate your customized story
-
-### Visual Storytelling
-1. Enter your story prompt
-2. Choose output language
-3. Select the number of scenes (1-5)
-4. Pick your preferred artistic style
-5. Generate a story with matching illustrations
-
-## Template System
-
-The application includes a sophisticated template system with:
 
-- **Adventure Templates**: Exploration and discovery narratives
-- **Friendship Templates**: Stories about connections and relationships
-- **Fantasy Templates**: Tales of magic and extraordinary powers
+### Content Safety System
+- Multi-layered content filtering
+- Regex pattern detection for inappropriate content
+- Repetition detection (unique word ratio < 0.4)
+- Fallback mechanisms to reliable templates
 
-Each template category includes multiple variations in both languages, ensuring fresh and engaging content each time. The system automatically:
+## Implementation Highlights
 
-1. Analyzes user prompts for keywords
-2. Selects the most appropriate template type
-3. Extracts parameters from the prompt when possible
-4. Uses default parameters when needed
-5. Customizes the selected template for a personalized story
+### MultilingualStoryGenerator Class
+- Central class managing generation in both languages
+- Handles language detection, content safety, and sentiment analysis
+- Template selection logic based on keyword matching
+- Parameter extraction from prompts
 
-## Safety Features
+### Story Templates
+- Three categories: Adventure, Friendship, Fantasy
+- Multiple variations in both languages
+- Dynamic parameter filling
 
-The application prioritizes content safety through:
+### GPU Resource Management
+- @spaces.GPU decorator for efficient GPU allocation
+- Pipeline moved to GPU only when needed for image generation
+- Proper cleanup with torch.cuda.empty_cache() and gc.collect()
 
-1. **Content Filtering**: Detection of inappropriate terms or patterns
-2. **Language Consistency**: Verification of output language integrity
-3. **Quality Control**: Detection of repetitive or nonsensical content
-4. **Fallback Mechanisms**: Multiple layers of backup generation options
+### Error Handling
+- Comprehensive logging system
+- Graceful degradation for missing components
+- Multiple fallback mechanisms
 
 ## Technical Requirements
-
 - Python 3.8+
-- CUDA-capable GPU recommended for image generation
-- Dependencies listed in requirements.txt
-
-## Future Enhancements
-
-- Enhanced Arabic image prompt understanding
-- Voice narration for stories
-- Interactive branching narratives
-- Additional language support
-- Expanded template library
-
-## License & Acknowledgements
-
-- [Hugging Face Transformers](https://github.com/huggingface/transformers)
-- [Diffusers](https://github.com/huggingface/diffusers)
-- [CAMeL-Lab](https://huggingface.co/CAMeL-Lab)
-- [Gradio](https://github.com/gradio-app/gradio)
-- [Helsinki-NLP](https://huggingface.co/Helsinki-NLP)
-
-## Contact
-
-For questions or support, please open an issue in the repository.
+- CUDA-capable GPU (for image generation)
+- Key dependencies: transformers, diffusers, gradio, torch