Madras1/Jade1.7b

Hugging Face
TEXT GENERATIONPricing:Input $0.32 / Cached $0.064 / Output $1.6Concurrent Unit Cost:1Model Size:2BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Mar 24, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Warm

Jade1.7b is a 1.7 billion parameter conversational finetune of Qwen3 developed by Madras1, specifically designed for Brazilian Portuguese chat. This model excels at expressing a strong, persistent persona with colloquial language, abbreviations, and a WhatsApp-like tone. It is optimized for PT-BR chat, chatbot use cases, and character-style interaction, making it ideal for informal conversational agents.

Loading preview...

Overview

Jade1.7b is a 1.7 billion parameter conversational model developed by Madras1, fine-tuned from unsloth/qwen3-1.7b. Its primary distinction is its persona-first design, meaning the model inherently speaks with a consistent "Jade" personality in Brazilian Portuguese (pt-BR) without requiring extensive system prompts. This results in informal phrasing, slang, and a friendly, chat-oriented tone from the first interaction.

Key Capabilities & Characteristics

  • Consistent Persona: Maintains a strong, informal, and friendly persona in Brazilian Portuguese.
  • Colloquial Language: Naturally uses abbreviations (e.g., vc), light slang, and short expressions (e.g., tmj, mano).
  • Chat-Optimized: Designed for casual, human-like conversations, mimicking a WhatsApp-style interaction.
  • Persona in Weights: The finetuning objective was to embed the persona directly into the model's weights, making it less reliant on explicit prompting for character consistency.

Training Details

The model was trained on approximately 25,000 examples over 3 epochs using an Unsloth-based SFT pipeline. The training involved synthetic PT-BR prompt generation and persona-driven response distillation, with system persona instructions removed during SFT to internalize the Jade style.

Recommended Use Cases

  • PT-BR Chat Assistants: Ideal for creating engaging and informal conversational agents.
  • Persona Bots: Excellent for applications requiring a consistent character or personality.
  • WhatsApp-style Agents: Suited for lightweight entertainment or social AI experiences that mimic casual messaging.

Limitations

Due to its strong persona orientation, Jade1.7b may not be suitable for:

  • Formal writing or contexts requiring a neutral tone.
  • High-stakes applications such as legal, medical, or financial advice, where strict formality and neutrality are critical.

Finetunes
1 models