json-schema-spec VS outlines

Yep and that comes from JSON Schema: https://json-schema.org/

I believe recent versions of OpenAPI are "compatible" with JSON Schema (at least they "wanted to be" last I checked as I was implementing some schema converters).

Even TypeScript is not enough to represent all of JSON Schema! But it gets close (perhaps if you remove validation rules and stuff like that it's a full match).

But even something like Java can represent most of it pretty well, specially since sealed interfaces were added. I know because I've done it :).

models.tf where I centralized all the Data model that API Gateway uses to perform input and output checks. Those use the JSON-schema specification. GitHub - psantus/serverless.api-gateway-dynamodb-integration.terraform

In the meantime, we are going to expand our backend with two endpoints: one for fetching data and another one for creating data. Fastify provides out-of-the-box support for API serialization and validation through its schema-based approach built on top of JSON Schema. Through the schema option, we can attach a schema definition to each route.

JSON-schema to define templates for request and response contents.

The syntax used to define the output is called JSON Schema. It is a standard way to define the structure of a JSON object. If you know zod, the spirit is similar. Based on Swarmion's roadmap, it will be possible to use zod schemas to defined contracts in the future, which will be super cool!

Essentially, all it does is attempt to generate the parameters to hypothetical or potential functions, which you using a JSON schema describe to ChatGPT.

[4] https://github.com/outlines-dev/outlines/tree/main

The next step is defining how to guide generation. For this step, we'll use the Outlines library. Outlines is a library for controlling how tokens are generated. It applies logic to enforce schemas, regular expressions and/or specific output formats such as JSON.

No benchmarks, just my anecdotal experience trying to get local LLM's to respond with JSON. The method above works for my use case nearly 100% of the time. Other things I've tried (e.g. `outlines`[0]) are really slow or don't work at all. Would love to hear what others have tried!

0 - https://github.com/outlines-dev/outlines

As I was playing with the Outlines library (https://outlines-dev.github.io/outlines/), I discussed with my friend Maxime how funny it would be if we set up a way to pair LLMs in chess matches till one wins. The first time I tried it, it required substantial prompt engineering to get some of those LLMs to propose valid moves. Large language models can mostly stay focused and even play rather well; see https://news.ycombinator.com/item?id=37616170 for example. However small language models aren't as easy to convince.

Some of those LLMs have seen very little chess notation and so after the first few opening moves there aren't any valid tactics, let alone strategy, so they would end up either repeating the same move, or hallucinate moves that are not valid (Kxe5, but there would be a queen on e5!)

Then Outlines came along and we could force them to pick valid moves with little cost! Maxime worked super fast and got a first version of this idea as a gradio space.

I think it is pretty fun to see the (mostly terrible, but otherwise valid) chess that those LLMs play. Maybe it will even be instructive to how we can create small LLMs that can play much better than the ones on the leaderboard.

Anyway, you can check it out here:

https://huggingface.co/spaces/mlabonne/chessllm

What is interactive about it: you can pick the LLMs from available models on HuggingFace (within reason, small LLMs are preferable so that the space does not crash) or push one of your own small models to HF and have it fight with others. At the end of the game the leaderboard is updated.

Hope you find it fun!

> The most obvious usage of this is forcing a model to output valid JSON

Isn't this something that Outlines [0], Guidance [1] and others [2] already solve much more elegantly?

0. https://github.com/outlines-dev/outlines

1. https://github.com/guidance-ai/guidance

2. https://github.com/sgl-project/sglang

And I think “just” Markdown doesn’t quite cut it for safe guidance. For example: directly generating content for your components. But I’m really excited about tooling like outlines appearing, with a greater focus on guided generation for structured data. Because this is often what we actually need!

It's pretty easy to force a locally running model to always output valid JSON: when it gives you probabilities for the next tokens, discard all tokens that would result in invalid JSON at that point (basically reverse parsing), and then apply the usual techniques to pick the completion only from the remaining tokens. You can even validate against a JSON schema that way, so long as it is simple enough.

There are a bunch of libraries for this already, e.g.: https://github.com/outlines-dev/outlines

https://github.com/outlines-dev/outlines/blob/7fae436345e621... squares with my experience using LLMs for anything real

  sequence = generator("Alice had 4 apples and Bob ate 2. Write an expression for Alice's apples:")

There are "smaller" models, for example tinyllama 1.1B (tiny seems like an exaggeration). PHI2 is 2.7B parameters. I can't name a 500M parameter model but there is probably one.

The problem is they are all still broadly trained and so they end up being Jack of all trades master of none. You'd have to fine tune them if you want them good at some narrow task and other than code completion I don't know that anyone has done that.

If you want to generate json or other structured output, there is Outlines https://github.com/outlines-dev/outlines that constrains the output to match a regex so it guarantees e.g. the model will generate a valid API call, although it could still be nonsense if the model doesn't understand, it will just match the regex. There are other similar tools around.