IqbalHere/Meme-Karigar

Meme Karigar

stars

commits

Python

primary language

Jul 11, 2026

updated

README

A fast Python translation pipeline for converting English text into Indic languages using IndicTrans2 with robust preprocessing and postprocessing.

🚀 Quick Start • 📖 Docs • 🔧 Config • 🤝 Contributing

✨ Why Meme Karigar?

🌍	English → Indic translation Uses `ai4bharat/indictrans2-en-indic-dist-200M` for multilingual translation from English to Indic scripts.
⚙️	Inference-ready text pipeline Preprocesses punctuation, tokenization, and language tags before generation.
🛡️	Placeholder preservation Protects URLs, emails, numerals, and tagged entities during translation and restores them after decoding.
🚀	GPU/CPU adaptive runtime Automatically picks CUDA when available and falls back to CPU otherwise.
🧩	Reusable API function Simple `translate_text(...)` interface for scripts and downstream integrations.

🏗️ Architecture

graph LR
    A[📝 Input Text] --> B[🧹 IndicProcessor Preprocess]
    B --> C[🔤 Tokenizer]
    C --> D[🤖 IndicTrans2 Model]
    D --> E[📤 Decode + Postprocess]
    E --> F[✅ Translated Output]

    style A fill:#6C63FF,color:#fff,stroke:#4C46B8
    style B fill:#A855F7,color:#fff,stroke:#7E22CE
    style C fill:#F59E0B,color:#111,stroke:#B45309
    style D fill:#10B981,color:#fff,stroke:#047857
    style E fill:#EF4444,color:#fff,stroke:#B91C1C
    style F fill:#6C63FF,color:#fff,stroke:#4C46B8

📦 Prerequisites

Requirement	Badge	Notes
Python		Runtime for scripts
PyTorch		Model execution backend
Transformers		Model and tokenizer loading

🚀 Quick Start

1️⃣ Clone repository

git clone https://github.com/IqbalHere/Meme-Karigar.git
cd Meme-Karigar

2️⃣ Create and activate virtual environment

macOS/Linux

python -m venv .venv
source .venv/bin/activate

Windows

python -m venv .venv
.venv\Scripts\activate

3️⃣ Install dependencies

pip install torch transformers indic-nlp-library sacremoses regex tqdm

4️⃣ Run sample translation

python /home/runner/work/Meme-Karigar/Meme-Karigar/test_translation.py

[!TIP] First run may take longer because model weights are downloaded from Hugging Face.

🔧 Configuration

Variable	Required	Description
`src_lang`	✅	Source FLORES code passed to `translate_text` (default: `eng_Latn`)
`tgt_lang`	✅	Target FLORES code passed to `translate_text` (example: `tel_Telu`)
`text`	✅	Input string to translate

[!NOTE] This repository currently uses function parameters for configuration; no environment variable is required by the checked-in Python code.

📚 API / Tool Reference

Parameter	Type	Required	Default	Description
`text`	`str`	✅	—	Input sentence/text
`src_lang`	`str`	❌	`eng_Latn`	Source language FLORES code
`tgt_lang`	`str`	❌	`tel_Telu`	Target language FLORES code

Example structured output:

{
  "translated_text": "ఇది పరీక్ష.",
  "source_language": "eng_Latn",
  "target_language": "tel_Telu"
}

🛡️ Error Handling

Scenario	Behavior
⚠️ Unsupported FLORES mapping	Falls back to default ISO behavior where applicable in `IndicProcessor` (`hi`)
❌ Model/tokenizer load issue	Raises dependency/runtime exception from Transformers/PyTorch
🔌 No CUDA available	Automatically runs on CPU

🧰 Development

Dev commands

# Run translation smoke test
python /home/runner/work/Meme-Karigar/Meme-Karigar/test_translation.py

Project structure

📦 Meme-Karigar
├── 📄 README.md
├── 📄 translator.py
├── 📄 indic_processor.py
├── 📄 test_translation.py
├── 📄 output.txt
└── 📄 useme.txt

🛠️ Tech Stack

Technology	Role
	Core implementation language
	Tensor runtime for inference
	Pretrained model/tokenizer APIs
	Indic normalization/tokenization/transliteration
	English punctuation normalization/tokenization

🤝 Contributing

Fork this repository
Create a branch (git checkout -b feature/your-change)
Commit your work (git commit -m "feat: describe your change")
Push your branch (git push origin feature/your-change)
Open a Pull Request

📜 License

License file is not currently present in this repository.

[!IMPORTANT] Add a /LICENSE file to clearly define usage rights for contributors and users.

Built with ❤️ by Iqbal

⭐ Star this repo if you found it useful! ⭐

Contributors

IqbalHere

6 commits

Copilot

1 commits

Similar repos

in-rolls/indicate

transliterate hindi, punjabi to english

AI4Bharat/setu-translate

RealImage/labs-aitt-ai4b

This repository contains tools for Indian language processing: IndicTrans for script…

VivekNeer/LangTranslation

neha13rana/Guj-Eng-Translator-

PALLADIUM26/translaTHOR

A language translation project

AI4Bharat/IndicTrans2

Translation models for 22 scheduled languages of India

470

ai4bharat/indictrans2-en-indic-dist-200M

Model

Languages

Python

100.0%

IqbalHere/Meme-Karigar

Meme Karigar

stars

commits

Python

primary language

Jul 11, 2026

updated

README

A fast Python translation pipeline for converting English text into Indic languages using IndicTrans2 with robust preprocessing and postprocessing.

🚀 Quick Start • 📖 Docs • 🔧 Config • 🤝 Contributing

✨ Why Meme Karigar?

🌍	English → Indic translation Uses `ai4bharat/indictrans2-en-indic-dist-200M` for multilingual translation from English to Indic scripts.
⚙️	Inference-ready text pipeline Preprocesses punctuation, tokenization, and language tags before generation.
🛡️	Placeholder preservation Protects URLs, emails, numerals, and tagged entities during translation and restores them after decoding.
🚀	GPU/CPU adaptive runtime Automatically picks CUDA when available and falls back to CPU otherwise.
🧩	Reusable API function Simple `translate_text(...)` interface for scripts and downstream integrations.

🏗️ Architecture

graph LR
    A[📝 Input Text] --> B[🧹 IndicProcessor Preprocess]
    B --> C[🔤 Tokenizer]
    C --> D[🤖 IndicTrans2 Model]
    D --> E[📤 Decode + Postprocess]
    E --> F[✅ Translated Output]

    style A fill:#6C63FF,color:#fff,stroke:#4C46B8
    style B fill:#A855F7,color:#fff,stroke:#7E22CE
    style C fill:#F59E0B,color:#111,stroke:#B45309
    style D fill:#10B981,color:#fff,stroke:#047857
    style E fill:#EF4444,color:#fff,stroke:#B91C1C
    style F fill:#6C63FF,color:#fff,stroke:#4C46B8

📦 Prerequisites

Requirement	Badge	Notes
Python		Runtime for scripts
PyTorch		Model execution backend
Transformers		Model and tokenizer loading

🚀 Quick Start

1️⃣ Clone repository

git clone https://github.com/IqbalHere/Meme-Karigar.git
cd Meme-Karigar

2️⃣ Create and activate virtual environment

macOS/Linux

python -m venv .venv
source .venv/bin/activate

Windows

python -m venv .venv
.venv\Scripts\activate

3️⃣ Install dependencies

pip install torch transformers indic-nlp-library sacremoses regex tqdm

4️⃣ Run sample translation

python /home/runner/work/Meme-Karigar/Meme-Karigar/test_translation.py

[!TIP] First run may take longer because model weights are downloaded from Hugging Face.

🔧 Configuration

Variable	Required	Description
`src_lang`	✅	Source FLORES code passed to `translate_text` (default: `eng_Latn`)
`tgt_lang`	✅	Target FLORES code passed to `translate_text` (example: `tel_Telu`)
`text`	✅	Input string to translate

[!NOTE] This repository currently uses function parameters for configuration; no environment variable is required by the checked-in Python code.

📚 API / Tool Reference

Parameter	Type	Required	Default	Description
`text`	`str`	✅	—	Input sentence/text
`src_lang`	`str`	❌	`eng_Latn`	Source language FLORES code
`tgt_lang`	`str`	❌	`tel_Telu`	Target language FLORES code

Example structured output:

{
  "translated_text": "ఇది పరీక్ష.",
  "source_language": "eng_Latn",
  "target_language": "tel_Telu"
}

🛡️ Error Handling

Scenario	Behavior
⚠️ Unsupported FLORES mapping	Falls back to default ISO behavior where applicable in `IndicProcessor` (`hi`)
❌ Model/tokenizer load issue	Raises dependency/runtime exception from Transformers/PyTorch
🔌 No CUDA available	Automatically runs on CPU

🧰 Development

Dev commands

# Run translation smoke test
python /home/runner/work/Meme-Karigar/Meme-Karigar/test_translation.py

Project structure

📦 Meme-Karigar
├── 📄 README.md
├── 📄 translator.py
├── 📄 indic_processor.py
├── 📄 test_translation.py
├── 📄 output.txt
└── 📄 useme.txt

🛠️ Tech Stack

Technology	Role
	Core implementation language
	Tensor runtime for inference
	Pretrained model/tokenizer APIs
	Indic normalization/tokenization/transliteration
	English punctuation normalization/tokenization

🤝 Contributing

Fork this repository
Create a branch (git checkout -b feature/your-change)
Commit your work (git commit -m "feat: describe your change")
Push your branch (git push origin feature/your-change)
Open a Pull Request

📜 License

License file is not currently present in this repository.

[!IMPORTANT] Add a /LICENSE file to clearly define usage rights for contributors and users.

Built with ❤️ by Iqbal

⭐ Star this repo if you found it useful! ⭐

Similar repos

in-rolls/indicate

transliterate hindi, punjabi to english

AI4Bharat/setu-translate

RealImage/labs-aitt-ai4b

This repository contains tools for Indian language processing: IndicTrans for script…

VivekNeer/LangTranslation

neha13rana/Guj-Eng-Translator-

PALLADIUM26/translaTHOR

A language translation project

AI4Bharat/IndicTrans2

Translation models for 22 scheduled languages of India

470

ai4bharat/indictrans2-en-indic-dist-200M

Model

Contributors

IqbalHere

6 commits

Copilot

1 commits

Languages

Python

100.0%