dphn/Dolphin3.0-Qwen2.5-1.5B

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:1.5BQuant:BF16Ctx Length:32kPublished:Jan 2, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Warm

Dolphin3.0-Qwen2.5-1.5B is a 1.5 billion parameter instruct-tuned model developed by Eric Hartford, Ben Gitter, BlouseJury, and Cognitive Computations, built upon the Qwen 2.5 architecture. Designed for general-purpose local use, it supports a 131072-token context length and excels in coding, math, agentic tasks, and function calling. This model prioritizes user control over system prompts and alignment, offering a steerable alternative to proprietary LLMs.

Loading preview...

Dolphin 3.0 Qwen 2.5 1.5B Overview

Dolphin 3.0 Qwen 2.5 1.5B is part of the Dolphin 3.0 series, developed by Eric Hartford, Ben Gitter, BlouseJury, and Cognitive Computations. This 1.5 billion parameter model, based on the Qwen 2.5 architecture, is instruct-tuned for a wide range of applications, including coding, mathematics, agentic behaviors, and function calling. A key differentiator is its emphasis on user control, allowing developers to define system prompts and alignment without external interference, addressing common issues with proprietary models like system prompt control, model version stability, and data privacy.

Key Capabilities & Features

  • General Purpose: Designed for broad utility across various tasks.
  • User Steerability: Offers full control over system prompts and alignment, enabling custom ethical guidelines and behaviors.
  • Data Privacy: Keeps user data local, preventing external access or use.
  • Long Context: Supports a context length of 131072 tokens.
  • ChatML Template: Utilizes the ChatML format for chat interactions.

Ideal Use Cases

  • Local AI Applications: Suitable for scenarios requiring a powerful, locally run model.
  • Customizable AI Agents: Excellent for building agents where specific behaviors and ethical frameworks are paramount.
  • Coding & Math Assistance: Strong performance in programming and mathematical problem-solving.
  • Function Calling: Capable of handling function calling tasks effectively.
  • Privacy-Sensitive Applications: Preferred for use cases where data privacy and control are critical.

Finetunes
1 models