Updates to Azure AI, Phi 3 tuning, and Gen AI models

Introducing new generative AI models, Phi 3 fine tuning, and other Azure AI enhancements to enable businesses to scale and personalise AI applications

[{"selector":"#anim-4e0d4189-fad1-437f-bbb5-39f68bb4a95a [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(34.249999904429615%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-e90389f8-a59c-4b2a-b1af-ca330b0c0384","keyframes":{"transform":["translate3d(115.50632%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-07889b1d-6ec2-4a2f-9f98-d7a8f6d5c57d","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-31cb924a-3f4e-485e-95ff-34c19e0bbbeb","keyframes":{"transform":["scale(0.15)","scale(1)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"forwards"}]

All sectors are being transformed by artificial intelligence, which also creates fresh growth and innovation opportunities

[{"selector":"#anim-9be8681e-2c2a-43a0-8327-5d59320cced3 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(34.249999904429615%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-0c892318-071b-4f5e-a8bd-b059c467bff2","keyframes":{"transform":["rotate(-540deg) scale(0.1)","none"],"opacity":[0,1]},"delay":0,"duration":1000,"fill":"both","iterations":1}]

Developers can rapidly and simply customise the Phi-3-mini and Phi-3-medium models for cloud and edge scenarios with serverless fine-tuning

[{"selector":"#anim-33e543da-18ad-450d-b353-9267e4275a47 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(35.57921965771819%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-b6118cc4-e6ad-4eb2-9108-b2e81a18900a","keyframes":{"transform":["translate3d(-119.93671%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-b6811715-b599-4957-a40d-48e2437d426c","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-41754352-26bc-4af7-b967-554eb54f5bb4","keyframes":{"transform":["scale(0.15)","scale(1)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"forwards"}]

These updates include a considerable improvement in core quality, instruction-following, and organised output

[{"selector":"#anim-03d7fd29-da32-4441-894d-f89abd0bbd2e [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(-34.249999904429615%, 0, 0) translate(-25%, 0%) scale(1.5)","translate3d(0%, 0, 0) translate(0%, 0%) scale(1)"]},"delay":0,"duration":2000,"fill":"forwards"}] [{"selector":"#anim-aa05f4c8-0ad1-4f94-88ff-4b7558921202","keyframes":{"transform":["translate3d(-117.15211%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-e11953ab-c6c1-4bac-8139-0cd78f8c6cf2","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-7b36fc08-ee34-4dc2-a21a-a51f0381b728","keyframes":{"transform":["scale(0.15)","scale(1)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"forwards"}]

Phi-3 models are well suited for fine-tuning to improve base model performance across a variety of scenarios

[{"selector":"#anim-7cf88bd4-cc5c-4c20-b409-e3f46a414c48 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate(0%, 0%) scale(1.5)","translate(0%, 0%) scale(1)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"forwards"}] [{"selector":"#anim-98f786b6-c05c-482a-ac43-63bbaf841121","keyframes":{"transform":["translate3d(115.18987%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-dd592238-9669-4424-aefb-94ea0abeb3ed","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-9ef0a370-d18e-4d1a-bf28-3ed9c6febc61","keyframes":{"transform":["scale(0.15)","scale(1)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"forwards"}]

A study from Khan Academy, which includes benchmarks from an improved version of Phi-3

[{"selector":"#anim-a701b247-87c6-4829-a749-c341f96aa50b [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(34.30232548614247%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-d29d0996-12d1-4f2c-b3de-9e645289ed76","keyframes":{"transform":["translate3d(115.78947%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-f6f0d6c6-4c6d-48e2-ae40-fa36e5c421d1","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-8268c6d0-2888-49a3-a381-87297dda17a2","keyframes":{"transform":["scale(0.15)","scale(1)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"forwards"}]

The state-of-the-art short language model (SLM) for the Neural Processing Unit (NPU) and shipping inbox is exclusive to Microsoft Windows

[{"selector":"#anim-3e813fae-e1a6-4f2b-8399-961fe638da2d [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(-34.249999904429615%, 0, 0) translate(-25%, 0%) scale(1.5)","translate3d(0%, 0, 0) translate(0%, 0%) scale(1)"]},"delay":0,"duration":2000,"fill":"forwards"}] [{"selector":"#anim-625d0fd1-3b15-4f93-bbab-c754a5319207","keyframes":{"transform":["translate3d(115.18987%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-e174fea7-ab2b-4405-b318-317d2c9e9fcf","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-ea704e18-7d0b-4e61-888e-accd7d0e7181","keyframes":{"transform":["scale(0.15)","scale(1)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"forwards"}]

The multi-modal Phi-3 model, Phi-3-vision, was unveiled at Microsoft Build and may be accessed via the Azure AI model catalogue For more details govindhtech.com

[{"selector":"#anim-dd192c14-9b80-497c-9a93-aee1d6eb5417 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(34.30232548614247%, 0, 0)","translate3d(0%, 0, 0)"]},"delay":0,"duration":2000,"easing":"cubic-bezier(.3,0,.55,1)","fill":"both"}] [{"selector":"#anim-19bb5a72-8d18-4306-84c4-6a04f0fa5ab9","keyframes":{"transform":["translate3d(115.18987%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-d43261a4-85c9-4cc0-ad88-f99df4039a4a","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-793b37d8-167e-4088-a608-cd6ebd5fc0bd","keyframes":{"transform":["scale(0.15)","scale(1)"]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"forwards"}]