jondurbin/airoboros-l2-70b-gpt4-2.0

TEXT GENERATIONConcurrency Cost:4Model Size:69BQuant:FP8Ctx Length:32kPublished:Jul 30, 2023License:otherArchitecture:Transformer0.0K Cold

jondurbin/airoboros-l2-70b-gpt4-2.0 is a 69 billion parameter instruction-tuned Llama-2 model developed by jondurbin, utilizing synthetic instructions generated exclusively by GPT-4 (0614 version). This model is specifically fine-tuned for context-obedient question answering, advanced coding tasks, agent/function calling, and chain-of-thought reasoning. It excels at adhering to provided context to reduce hallucinations and supports complex execution planning.

Loading preview...

Overview

jondurbin/airoboros-l2-70b-gpt4-2.0 is a 69 billion parameter instruction-tuned Llama-2 model, fine-tuned using synthetic instructions generated by jondurbin's Airoboros project. This 2.0 series model is unique as its training data was exclusively generated by the 0614 version of GPT-4, aiming for a cleaner dataset. It's a QLoRA fine-tune, building on previous versions of this size.

Key Capabilities

  • Context-Obedient QA: Trained to strictly adhere to provided context, minimizing hallucinations by ignoring external knowledge.
  • Advanced Coding: Capable of handling complex coding instructions, including multi-criteria requests and generating plain code output.
  • Agent/Function Calling: Supports generating JSON or YAML outputs for function calls, similar to OpenAI's function calling.
  • Chain-of-Thought Reasoning: Can provide multiple potential answers, rank them by logic, and select the most feasible one.
  • reWOO-style Execution Planning: Generates systematic plans for complex instructions requiring multiple tool uses, outputting a sequence of function calls.

Good For

  • Applications requiring highly accurate, context-bound responses.
  • Developers needing a model for complex code generation or agentic workflows.
  • Use cases benefiting from structured reasoning and multi-step problem-solving.

Licensing Note

This model is based on Llama-2, subject to Meta's custom license. The fine-tuning data was generated via OpenAI API calls, leading to an ambiguous license regarding commercial use due to OpenAI's ToS. Users should exercise caution for commercial deployment.