- Dev Notes
- Posts
- OpenAI Introduces Structured Outputs for Reliable JSON in the API
OpenAI Introduces Structured Outputs for Reliable JSON in the API
Good Morning! OpenAI has introduced Structured Outputs, a new feature in their API that ensures model-generated JSON outputs will exactly match developer-provided schemas. Google's Pigweed integration has landed on the Raspberry Pi Pico 2, bringing powerful build tools and hardware-agnostic libraries to simplify embedded development.
OpenAI Introduces Structured Outputs for Reliable JSON in the API
OpenAI is now introducing Structured Outputs in the API - a new feature designed to ensure model-generated outputs will exactly match JSON Schemas provided by developers. This solves the problem of unreliable model outputs not conforming to the required formats.
Structured Outputs is available in two forms:
Function Calling: Developers can set
strict: true
within their function definition, and the model outputs will match the supplied tool schema.Response Format: Developers can now supply a JSON Schema via the
json_schema
option in theresponse_format
parameter. This is useful when the model is responding to the user directly, rather than calling a tool.
Benefits:
Guaranteed JSON outputs that match developer-supplied schemas
Native SDK support in Python and Node for easy integration
Maintains OpenAI's safety policies, allowing models to refuse unsafe requests
Availability: Structured Outputs is generally available now in the OpenAI API. The new feature is compatible with all models that support function calling, including the latest GPT-4 models.
In addition, OpenAI has released a new model, gpt-4o-2024-08-06
, which offers a 50% reduction in input token pricing and a 33% reduction in output token pricing compared to the previous gpt-4o-2024-05-13
model.
Read More Here
Plugged In
Rim or Pocket? Engineering Spikeball’s Biggest Problem
Now featuring our second article for PluggedIn! In case you don’t know, PluggedIn is our alternate non-profit venture of the people at DevNotes where we interview and write blogs about the brightest young tech minds around the world. Have you ever found yourself in a heated Spikeball dispute over whether the ball hit the rim or the pocket? If so, you're not alone. For this article we are writing about Evan Talley, a recent high school graduate and incoming mechanical engineering student at Northern Kentucky University, who took on this very problem for his senior capstone project. Despite facing tight budget constraints and technical hurdles, Evan and his team developed a sensor-based device to detect the difference between rim and pocket shots in Spikeball. Their journey showcases the real-world challenges of bringing an engineering concept to life, and their innovative solution could revolutionize the way we play this trendy sport. Watch Video
Read More Here
Google Pigweed Lands on the Raspberry Pi Pico 2
The Pigweed integration brings some truly exciting capabilities to the Pico 2. Developers can now take advantage of Bazel, a cutting-edge build system that simplifies the build and testing process for large, complex projects. With Bazel, you can enjoy hermetic, reproducible builds, a modern Clang/LLVM toolchain, and cross-platform support – all without the hassle of managing complex build systems.
Pigweed also provides a wealth of hardware-agnostic C++ libraries, making it easier to structure your codebase around sensible abstractions. Plus, you can leverage the Pigweed REPL to interactively communicate with your Pico, view logs, and send commands – perfect for rapid prototyping and debugging.
Getting Started: To get started with Pigweed on your Pico 2, simply head over to the Pigweed documentation and check out the Raspberry Pi Pico showcase. You'll find detailed tutorials, sample code, and a comprehensive demo that showcases the power of this integration.
Read More Here
🔥 More Notes
Youtube Spotlight
world's shortest Java course
Was this forwarded to you? Sign Up Here