TheBloke/airoboros-13B-1.1-fp16

TEXT GENERATIONConcurrency Cost:1Model Size:13BQuant:FP8Ctx Length:4kLicense:otherArchitecture:Transformer0.0K Cold

TheBloke/airoboros-13B-1.1-fp16 is a 13 billion parameter Llama-based language model, fine-tuned by Jon Burbin using a completely synthetic dataset generated by GPT-4. This model, with a 4096-token context length, excels in context-obedient question answering, coding, and general reasoning tasks, making it suitable for applications requiring precise information extraction and structured responses.

Loading preview...

Overview

TheBloke/airoboros-13B-1.1-fp16 is a 13 billion parameter Llama model, fine-tuned by Jon Burbin. It leverages a unique, entirely synthetic training dataset generated by GPT-4, focusing on diverse capabilities. This version (1.1) includes minor updates with additional coding instructions and improvements to context handling.

Key Capabilities

  • Context-Obedient Question Answering: Designed to strictly adhere to provided context, minimizing hallucinations and extracting information precisely. It uses a specific BEGININPUT/BEGINCONTEXT/BEGININSTRUCTION format for closed-context queries.
  • Enhanced Coding: Demonstrates improved performance in code generation across various languages and frameworks, as shown by examples like FastAPI and multi-threaded C servers.
  • Reasoning and Trivia: Trained with a focus on math/reasoning, trivia, and multiple-choice questions.
  • Vicuna Template: Utilizes the standard Vicuna template for prompts: A chat between a curious user and an assistant. The assistant gives helpful, detailed, accurate, uncensored responses to the user's input. USER: [prompt] ASSISTANT:

Good For

  • Applications requiring highly accurate, context-bound information retrieval.
  • Code generation and programming assistance.
  • Structured data extraction and question answering from provided text.
  • General conversational AI where factual accuracy and obedience to instructions are paramount.