DataPilot/ArrowMint-Gemma3-4B-YUKI-v0.1

VISIONPricing:Input $0.2 / Output $0.4Concurrent Unit Cost:1Model Size:4.3BQuant:BF16Context Size:32kPublished:Mar 29, 2025License:gemmaArchitecture:Transformer0.0K Featherless Exclusive Cold

DataPilot/ArrowMint-Gemma3-4B-YUKI-v0.1 is a 4.3 billion parameter Japanese language model based on Google's Gemma 3 architecture, fine-tuned for AI VTuber (AItuber) dialogue capabilities. It excels in multi-turn conversations, prompt adherence, and maintaining character consistency. This model is optimized for real-time, resource-efficient interaction, making it suitable for dynamic conversational AI applications.

Loading preview...

Overview

DataPilot/ArrowMint-Gemma3-4B-YUKI-v0.1 is a 4.3 billion parameter Japanese language model built upon Google's gemma-3-4b-it base model. Developed by DataPilot, this model is specifically fine-tuned for AItuber (AI VTuber) dialogue capabilities, focusing on natural and consistent interactions.

Key Capabilities

  • Multi-turn Performance: Designed to maintain coherent and natural dialogue flow over extended conversations.
  • Prompt Adherence: Highly responsive to user instructions and character settings, ensuring consistent and on-brand responses.
  • Lightweight and Efficient: Optimized for real-time response generation and operation in resource-constrained environments, crucial for live AItuber applications.
  • Japanese Language Focus: Specialized for high-quality interaction in Japanese.

Development and Architecture

The model was created using efficient fine-tuning techniques with Unsloth and synthetic datasets, followed by a mergekit process to combine the strengths of several specialized models. This approach allowed for the integration of various capabilities, including code, general Japanese, and instruction-following, into a single cohesive model.

Good For

  • AItuber Applications: Ideal for powering AI VTubers requiring sustained, character-consistent, and responsive dialogue.
  • Interactive Conversational AI: Suitable for scenarios demanding strong multi-turn capabilities and adherence to specific personas or instructions.
  • Japanese Language Processing: Excellent for applications where high-quality Japanese dialogue generation is paramount.

Finetunes
1 models