Orca-AgentInstruct: Agentic flows can be effective synthetic-data generators

Orca-3 blog - abstract wave graphic

Our work on Orca and Orca 2 demonstrated the power of using synthetic data for the post-training of small language models and getting them to levels of performance previously found only in much larger language models. Orca-AgentInstruct is another step in this direction, where we explore using agentic flows to generate diverse and high-quality data at scale. Orca-AgentInstruct is an agentic solution for synthetic-data generation. By leveraging an agentic framework, AgentInstruct can generate tailored datasets, comprising both prompts and responses, from raw data sources, paving the way to building a synthetic data factory for model fine-tuning.  

The efficacy of this approach is exemplified by the substantial improvement observed by fine-tuning a base Mistral 7-billion-parameter model and using AgentInstruct to generate a 25-million-pair dataset. The fine-tuned model (which we refer to as Orca-3-Mistral) showcases a notable performance gain across multiple benchmarks. For example, it shows 40% improvement on AGIEval, 19% improvement on MMLU, 54% improvement on GSM8K, 38% improvement on BBH, 45% improvement on AlpacaEval, and a 31.34% reduction of inaccurate or unreliable results across multiple summarization benchmarks.

We are making a 1-million-pair subset (orca-agentinstruct-1M) of this dataset publicly available, along with a report describing the data generation procedure, to encourage research on synthetic data generation and finetuning of language models. 

Bar graph comparing scores of the Mistral-Instruct-7B model and the Mistral-7B post-trained AgentInstruct data (Orca-3). The benchmarks are AGIEval, MMLU, BBH, GSM8K, AlpaceEval, FOFO and Mirage-RAG. The graph shows substantial improvement across different benchmarks for the model fine-tuned with AgentInstruct data.
Figure 1: Effect of using AgentInstruct for post-training Mistral-7B. 
The figure shows the three flows used in AgentInstruct: 1) Content Transformation Flow converts the raw seed into an intermediate representation that simplifies the creation of instructions tailored to specific objectives. 2) Seed Instruction Generation Flow, comprising multiple agents, takes as input the transformed seed from the Content Transformation Flow and generates a set of diverse instructions. 3) Instruction Refinement Flow takes as input the instructions from the Seed Instruction Flow and iteratively enhances their complexity and quality.
Figure 2. This figure provides a thematic overview of the roles played by different groups of agents. Content Transformation Flow converts the seed into an intermediate representation that makes it easier to create high-quality and diverse data. Seed Instruction Generation Flow creates instances of the target tasks following a taxonomy. Instruction Refinement Flow explores the space further by starting from these initial data points and exploring the neighborhood. The expectation is that by picking a random seed we will be able to cover the entire region of data points. 

Synthetic Data Accelerated LLM Development: Over the past year, using synthetic data has greatly advanced the training of large language models (LLMs). It sped up model training at all stages, from pre-training (e.g., Phi-3) to instruction-tuning (e.g., Orca and WizardLM) and reinforcement learning from human feedback (e.g., Direct Nash Optimization). 

Generating high-quality synthetic data is hard: On the other hand, research indicates that pre-training models on synthetic data produced by other models can result in model collapse, causing models to progressively degrade. Similar concerns have been raised regarding the use of synthetic data for post-training, suggesting that it might lead to an imitation process where the trained model learns only stylistic features rather than actual capabilities. 

This discrepancy may be attributed to the challenge of generating high-quality and diverse synthetic data.  Successful use of synthetic data involves significant human effort in curating and filtering the data to ensure high quality. 

Synthetic data meets agents: Another major development we witnessed during the past year is the rise of agentic (especially multi-agent) workflows, such as with AutoGen. Agentic workflows can generate high-quality data, which surpasses the capabilities of the underlying LLMs, by using flows with reflection and iteration that enable agents to look back at solutions, generate critiques, and improve solutions. They can also use tools like search APIs, calculators, and code interpreters to address LLM limitations. 

Multi-agent workflows bring in additional benefits as well, such as simulating scenarios where we can generate both new prompts and the corresponding responses. They also enable automation of data-generation workflows, reducing or eliminating the need for unnecessary human intervention on some tasks. 

AgentInstruct: Generating synthetic data for post-training or finetuning often relies on an existing prompt set that is either used as is or as seeds for generating more instructions. In this work, we generalize the problem settings to a broader objective of generating an abundant amount of diverse, challenging, and high-quality data to teach a particular skill to an AI model. We refer to this setting as generative teaching.   

AgentInstruct is an agentic solution for generative teaching. AgentInstruct uses raw documents as input to create demonstration and feedback data. When generic data is used as seeds, AgentInstruct can be used to teach an LLM a general capability, such as writing, reasoning, or retrieval-augmented generation (RAG). Domain specific data, like retail or finance, can also be used as seeds to improve the model in a certain specialization. AgentInstruct can create: 

  1. High-quality data: AgentInstruct uses GPT-4, coupled with tools like search and code interpreters, to create high-quality data.  
  2. Diverse data: AgentInstruct creates prompts and responses using a set of specialized agents (with powerful LLMs, tools, and reflection flows) and a taxonomy (of more than 100 subcategories), , ensuring diversity and quality.
  3. Large quantities of data: AgentInstruct can run autonomously. and applyiflows for verification and data filtering. It does not require seed prompts and uses raw documents for seeding. 

Using raw data as seeds offers two advantages: it is plentiful, allowing AgentInstruct to generate large-scale and diverse datasets, and it encourages learning general skills instead of benchmark-specific ones by avoiding using existing prompts.

Microsoft Research Blog

Microsoft Research Forum Episode 3: Globally inclusive and equitable AI, new use cases for AI, and more

In the latest episode of Microsoft Research Forum, researchers explored the importance of globally inclusive and equitable AI, shared updates on AutoGen and MatterGen, presented novel use cases for AI, including industrial applications and the potential of multimodal models to improve assistive technologies.


We anticipate agentic flows becoming increasingly important throughout the model-training lifecycle, including pre-training, post-training, and specialization, and ultimately enabling the creation of a synthetic data factory for model customization and continuous improvement. This has the potential to drive AI advances across multiple industries by making high-quality model training more efficient and accessible. 

Contributors:

Arindam Mitra, Luciano Del Corro, Guoqing Zheng, Shweti Mahajan, Dany Rouhana, Andres Codas, Yadong Lu, Wei-ge Chen, Olga Vrousgou, Corby Rosset, Fillipe Silva, Hamed Khanpour, Yash Lara, and Ahmed Awadallah

The post Orca-AgentInstruct: Agentic flows can be effective synthetic-data generators appeared first on Microsoft Research.

Read More