---

📌 Overview

Hermes is an advanced Flask-based AI translation and rephrasing web application powered by Groq's high-speed LLM models (using llama-3.1-8b-instant by default).

It dynamically adapts translations based on style, tone, dialect, and creativity—allowing you to render text in poetic forms, specific regional dialects, or preserve specific aspects of the original voice.

The application has been optimized to run both locally and as serverless functions on hosting providers like Vercel.

---

✨ Features

• Groq Cloud Integration — Ultra-fast inference leveraging the Groq API (replaces the old local Ollama setup).
• Style & Tone Control — Fine-tune output to match genres/styles like formal, casual, poetic, business, etc.
• Originality Preservation — Preserve specific dimensions of the input (meaning, tone, style, voice, form) during rephrasing.
• Dialect & Creative Intent — Direct the AI to translate using specific dialects (e.g. British English, Mexican Spanish) or apply custom creative rules.
• Serverless Ready — Pre-configured for seamless deployments on Vercel.
• Safe Client Initialization — Groq client is initialized lazily, preventing application crashes at startup when the API key is missing.

---

🌐 API Reference

POST /api/translate

Performs translation/rephrasing based on your parameters.

Headers

Content-Type: application/json

Request Body Options

Parameter	Type	Required	Default	Description
text	string	Yes	-	The source text to translate or rephrase.
target_lang	string	Yes	English	The target language for the output.
writing_style	string	No	-	Writing style/genre (e.g. casual, formal, business, poetic).
originality	string (JSON array)	No	[]	Attributes of the original text to preserve (e.g., ["meaning", "tone", "style", "voice", "form"]).
dialect	string	No	-	Specific regional dialect (e.g. Scottish, Brazilian).
creative_intent	string	No	-	Additional custom creative constraints or instructions.
model	string	No	llama-3.1-8b-instant	The Groq model to use.

Example Request

{
  "text": "Hello, how are you doing today?",
  "target_lang": "Spanish",
  "writing_style": "poetic",
  "originality": "[\"meaning\"]",
  "dialect": "Andalusion",
  "model": "llama-3.1-8b-instant"
}

Example Response

{
  "output": "Hola, ¿cómo te va en este día de luz?"
}

---