persona-chat 0.1.0

A sophisticated CLI application for generating authentic multi-turn conversations using 50+ distinct user personas

Persona-Based Conversation Generator CLI

A sophisticated Python CLI application that generates authentic multi-turn conversations using 50+ distinct user personas across multiple languages.

Features

• 50+ Personas: Technology, Healthcare, Education, Business, Creative, and Specialized domains
• Multi-Language Support: English, Arabic, and more
• Multi-Turn & Multi-Persona Conversations: Generate conversations with 3-20 turns, including multi-persona (role-based) dialogues
• Dynamic Personas: Personas are created on-the-fly; no static persona files required
• Batch Generation: Generate 1-1000 conversations simultaneously with random or domain-specific personas
• Flexible Export Formats: Output as JSON, CSV, Markdown, or plain text
• Quality Analysis: Built-in conversation quality metrics and validation
• Cultural Adaptation: Automatic cultural context and communication style adaptation

Quick Start

Installation

# Clone the repository
git clone <repository-url>
cd persona-chat

# Install dependencies
pip install -r requirements.txt

# Install the package
pip install -e .

Basic Usage

# Generate a single conversation with a random persona
persona-chat generate

# Use a specific persona with multiple turns
persona-chat generate --persona-name "Senior Software Engineer" --turns 8

# Generate an Arabic conversation
persona-chat generate --language ar --persona-name "مهندس برمجيات" --turns 6

# Batch generation (random personas)
persona-chat batch --conversations 10 --turns 5 --random-personas

# Batch generation (IT/technology personas)
persona-chat batch --conversations 10 --turns 5 --persona-domain technology

# Multi-persona conversation (role-based)
persona-chat generate --multi-persona --persona-names "Teacher,Student,Parent" --persona-roles "primary,participant,observer" --language ar --turns 4

Batch Generation Examples

Multi-Persona Batch Generation (NEW)

Batch now supports all persona options, including multi-persona, matching the generate command.

persona-chat batch --conversations 3 --turns 3 --language ar --multi-persona --persona-names "Teacher,Student,Parent" --persona-roles "primary,participant,observer" --llm-backend openai --llm-model gemma2:2b --llm-api-url http://localhost:11434/v1 --output json --output-filename test_multi_persona_batch

Arabic Batch with a Single Dynamic Technology Persona (All Conversations)

persona-chat batch --conversations 3 --turns 4 --language ar --create-persona --persona-name "مهندس برمجيات" --persona-role "مهندس برمجيات" --persona-domain technology --persona-expertise expert --persona-communication-style technical --llm-backend openai --llm-model gemma2:2b --llm-api-url http://localhost:11434/v1 --output json --output-filename test_arabic_batch_dynamic_it

Multi-Persona Conversation (Single)

persona-chat generate --multi-persona --persona-names "Chef,Waiter,Customer" --persona-roles "primary,participant,observer" --language ar --turns 3 --llm-backend openai --llm-model gemma2:2b --llm-api-url http://localhost:11434/v1 --output markdown --output-filename test_multi_persona_restaurant

Note: All persona options (including multi-persona, persona names, roles, and dynamic persona creation) are now supported in both batch and generate commands.

Troubleshooting:

• If you see No valid persona available for conversation generation, use the --create-persona and --persona-name options as shown above.
• If you want random personas per conversation, you must restore the static persona files or update the code to support dynamic random persona generation.

Export Formats

• --output json (default): Structured JSON
• --output csv: Tabular CSV
• --output markdown: Human-readable Markdown
• --output text: Plain text

LLM Backend Support (OpenAI, Ollama, Transformers)

You can use OpenAI-compatible APIs (including local Ollama) or HuggingFace Transformers as the backend for all message generation.

Configuration

• --llm-backend: Choose the backend (openai, transformers, or none)
• --llm-model: Model name (e.g., gemma2:2b for Ollama)
• --llm-api-key: (For OpenAI) Your OpenAI API key (use any dummy value for local Ollama)
• --llm-api-url: (For OpenAI) Custom API URL for self-hosted or alternative OpenAI-compatible APIs (e.g., http://localhost:11434/v1 for Ollama)

Example: Local Ollama API

persona-chat generate --llm-backend openai --llm-model gemma2:2b --llm-api-key dummy --llm-api-url http://localhost:11434/v1 --persona-name "Senior Software Engineer" --turns 5

Notes

• All user and assistant messages are generated using the selected LLM backend if enabled.
• For local Ollama, use --llm-backend openai and provide a dummy API key.
• For Transformers, the model must be available locally or downloadable from HuggingFace.
• You can set defaults in a config file or via environment variables for convenience.

Multi-Persona Conversations

• Use --multi-persona to enable multi-persona mode
• Specify persona names and roles with --persona-names and --persona-roles
• Example:

  persona-chat generate --multi-persona --persona-names "Artist,Designer,Client" --persona-roles "primary,participant,observer" --language ar --turns 3
  

Documentation

• Installation Guide
• Quick Start Guide
• User Guide
• Developer Guide
• Examples

Project Structure

persona-chat/
├── persona_chat/          # Main application package
├── config/               # Configuration files
├── output/               # Output files
├── tests/                # Test suite
├── docs/                 # Documentation
├── examples/             # Example files
└── tools/                # Development tools

Contributing

Please read CONTRIBUTING.md for details on our code of conduct and the process for submitting pull requests.

License

This project is licensed under the MIT License - see the LICENSE file for details.