Nucleus is a 4 billion parameter AI assistant developed by venerr, built upon the Qwen3-4B architecture. This model is fine-tuned for natural conversation across 14 languages, excelling in coding assistance, text tasks like summarization and translation, and providing witty, humorous responses. Its lightweight design and multilingual capabilities make it suitable for diverse conversational AI applications requiring efficiency and broad language support.
Loading preview...
Nucleus: A Smart, Multilingual AI Assistant
venerr/nucleus is a 4 billion parameter AI assistant, fine-tuned from the Qwen3-4B base model. It is designed for natural, engaging conversations, offering a blend of helpfulness and humor. The model is particularly notable for its broad language support and efficiency.
Key Capabilities
- Multilingual Conversation: Supports natural interaction in 14 languages, including Russian, English, Spanish, French, German, Chinese, Japanese, Korean, Portuguese, Italian, Polish, Arabic, Turkish, and Hindi.
- Coding Assistance: Provides help with various programming languages such as Python, SQL, and JavaScript.
- Text Processing: Capable of rewriting, translation, summarization, and generating essays.
- Engaging Responses: Delivers smart and witty replies, aiming for a friendly and humorous tone.
- Lightweight Design: With only 4 billion parameters, it offers fast performance suitable for efficient deployment.
Training Details
Nucleus was fine-tuned using QLoRA (LoRA rank 16) over 33 training steps and 3 epochs, utilizing a dataset of 86 examples. The training was performed with Unsloth + TRL framework.
Good For
- Applications requiring a conversational AI with broad language support.
- Integrating a helpful assistant for coding and text-based tasks.
- Use cases where a lightweight yet capable model is preferred.
- Creating interactive experiences with a touch of humor.