Dolphin3.0 Llama3.2 1 B

Introduction

Dolphin 3.0 is part of the Dolphin series, a collection of instruct-tuned models designed for coding, math, agentic tasks, function calling, and general use cases. Dolphin 3.0 aims to provide businesses with greater control and customization options compared to other AI models like ChatGPT or Claude.

Architecture

• Base Model: Meta-Llama/Llama-3.2-1B
• Language: English
• Datasets Used:
  • OpenCoder-LLM/opc-sft-stage1, opc-sft-stage2
  • Microsoft/orca-agentinstruct-1M-v1, orca-math-word-problems-200k
  • NousResearch/hermes-function-calling-v1
  • AI-MO/NuminaMath-CoT, NuminaMath-TIR
  • AllenAI/tulu-3-sft-mixture
  • CognitiveComputations/dolphin-coder, samantha-data
  • HuggingFaceTB/smoltalk
  • m-a-p/CodeFeedback-Filtered-Instruction, Code-Feedback

Training

The model was trained using resources provided by sponsors such as Crusoe Cloud, Akash, Lazarus, and Cerebras. Key features include providing customizable system prompts to set tone and alignment for the AI's responses.

Guide: Running Locally

1. Setup Environment:

   • Install Python and the Hugging Face Transformers library.
   • Set up a virtual environment and install necessary dependencies.

2. Download Model:

   • Retrieve the model from Hugging Face's model hub.

3. Run Model:

   • Use the model through various interfaces like Ollama, LM Studio, or directly via the Hugging Face Transformers library.

4. Cloud GPUs:

   • For enhanced performance, consider using cloud GPUs from providers like AWS, Google Cloud, or Azure.

License

The model is licensed under Llama3.2. For further details, refer to the licensing terms provided by the creators.