Microsoft Azure AI Foundry already helps mannequin fine-tuning. At the moment, Microsoft introduced main enhancements to mannequin fine-tuning on AI Foundry with assist for Reinforcement Fantastic-Tuning (RFT) and extra. RFT is a brand new approach that makes use of chain-of-thought reasoning and task-specific grading to enhance mannequin efficiency for particular domains.

OpenAI first introduced the alpha program for RFT final December. In response to their early testers, RFT delivered a 40% improve in mannequin efficiency when in comparison with commonplace out-of-the-box fashions. Microsoft right now introduced that RFT for OpenAI’s o4-mini mannequin is coming quickly to Azure AI Foundry. The corporate recommends RFT within the following situations:

• Customized Rule Implementation: RFT thrives in environments the place determination logic is very particular to your group and can’t be simply captured by means of static prompts or conventional coaching knowledge. It allows fashions to be taught versatile, evolving guidelines that mirror real-world complexity.
• Area-Particular Operational Requirements: Splendid for situations the place inside procedures diverge from trade norms—and the place success will depend on adhering to these bespoke requirements. RFT can successfully encode procedural variations, akin to prolonged timelines or modified compliance thresholds, into the mannequin’s conduct.
• Excessive Determination-Making Complexity: RFT excels in domains with layered logic and variable-rich determination timber. When outcomes rely upon navigating quite a few subcases or dynamically weighing a number of inputs, RFT helps fashions generalize throughout complexity and ship extra constant, correct choices.

Microsoft right now additionally introduced assist for Supervised Fantastic-Tuning (SFT) for OpenAI’s newest GPT-4.1-nano mannequin, which is appropriate for cost-sensitive AI purposes. Fantastic-tuning for GPT-4.1 will probably be accessible within the coming days.

Lastly, Microsoft introduced assist for fine-tuning Meta’s newest Llama 4 Scout 17 billion parameter mannequin that comes with assist for a 10M token context window. Additionally, the Llama 4 fine-tuning is accessible as a part of Azure’s managed compute providing. The fine-tuned Llama mannequin is accessible in each Azure AI Foundry and as Azure Machine Studying elements.