Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A guidance language for controlling LLMs

github.com

Post reply on HN
Print view View original

81–90 of 198 posts

Re: A guidance language for controlling LLMs

#81

Post by eternalban »

ubj wrote: post #38

I like this step towards greater rigor when working with LLM's. But part of me can't help but feel like this is essentially reinventing the concept of programming languages: formal and precise syntax to perform specific tasks with guarantees. I wonder where the final balance will end up between the ease and flexibility of everyday language, and the precision / guarantees of a formally specified language.

So far it it reminds of the worst days of code embedded in templates. Once these things start getting into multipage prompts they will be hopelessly obscure. The second immediate thing that jumps out is 'fragility'. This will be the sort of codebase that original "prompt engineer" wrote and left and no one will touch it for fear of breaking humpty dumpty.

Re: A guidance language for controlling LLMs

#82

Post by madrox »

ubj wrote: post #38

I like this step towards greater rigor when working with LLM's. But part of me can't help but feel like this is essentially reinventing the concept of programming languages: formal and precise syntax to perform specific tasks with guarantees. I wonder where the final balance will end up between the ease and flexibility of everyday language, and the precision / guarantees of a formally specified language.

The lovely thing about LLMs is that it can handle poorly worded prompts and well worded prompts. On the engineering side, we'll certainly see more rigor and best practices. For your average user? They can keep throwing whatever they like at it.

Re: A guidance language for controlling LLMs

#83

Post by leroy-is-here »

ryanklee wrote: post #70
leroy-is-here wrote: post #42

Earlier quoted context omitted.

Not dubious, I just read your comment and it felt like I was reading satire. Even the cadence of your words felt funny. Anyway, I’m not surprised. It’s a new market, everyone’s in on it.

You should have led with generosity instead of tacking it on at the end. It might have saved me from having a ridiculous conversation about the cadence of my words, and instead there might have been a higher chance of someone saying something substantive about my assumptions regarding the technology. But here we are.

I agree, I came off a tad harsh. Sorry about that

Re: A guidance language for controlling LLMs

#84

Post by behnamoh »

ryanklee wrote: post #11

I'm personally starting with learning Guidance and LMQL rather than LangChain just in order to get a better grasp of the behaviors that I've gathered LangChain papers over. Even after that, I'm likely to look at Haystack before LangChain. Just getting the feeling that LangChain is going to end up being considered a kitchen sink solution full of anti patterns so might as well spend time a little lower level while I se…

What I didn't like about langchain is the lack of consistent directories and paths for things.

Re: A guidance language for controlling LLMs

#85

Post by efitz »

jameshart wrote: post #43
ryanklee wrote: post #39

Earlier quoted context omitted.

Not quite sure what the spirit of your comment is. But, yes, they are real technologies. Very confused as to why you would even find that dubious.

Consider how similar your comment reads, for an outsider, to this explanation of AWS InfiniDash: https://twitter.com/TartanLlama/status/1410959645238308866

Hahaha "The first step of Byzantine Fault Tolerance is tolerance" omg. That cracked me up. Reminded me of the Rockwell Encabulator: https://youtu.be/RXJKdh1KZ0w

Re: A guidance language for controlling LLMs

#87

Post by PeterisP »

simonw wrote: post #21

This is pretty fascinating, but I'm not sure I understand the benefit of using a Handlebars-like DSL here. For example, given this code from https://github.com/microsoft/guidance/blob/main/notebooks/ch... create_plan = guidance('''{{#system~}} You are a helpful assistant. {{~/system}} {{#block hidden=True}} {{#user~}} I want to {{goal}}. {{~! generate potential options ~}} Can you please generate one option for how t…

Your example assumes a nested, hierarchical structure while the former example is strictly linear. IMHO that's the key difference there, as the former can (and AFAIK is) be directly encoded and passed to the LLM, which inherently receives only a flat list of tokens.

Your example might be nicer to edit, but then it would still have to be translated to the actual 'guidance language' which would have to look (and be) flat.

Re: A guidance language for controlling LLMs

#89

Post by startupsfail »

ubj wrote: post #38

I like this step towards greater rigor when working with LLM's. But part of me can't help but feel like this is essentially reinventing the concept of programming languages: formal and precise syntax to perform specific tasks with guarantees. I wonder where the final balance will end up between the ease and flexibility of everyday language, and the precision / guarantees of a formally specified language.

We really need to start thinking of how to reduce magical thinking in the field. It’s not pretty. They literally quote biblical guidance for the models and pray that this would work.

And start their prompts with “You”. Who is “You”?

Re: A guidance language for controlling LLMs

#90

Post by jweir »

madrox wrote: post #82
ubj wrote: post #38

I like this step towards greater rigor when working with LLM's. But part of me can't help but feel like this is essentially reinventing the concept of programming languages: formal and precise syntax to perform specific tasks with guarantees. I wonder where the final balance will end up between the ease and flexibility of everyday language, and the precision / guarantees of a formally specified language.

The lovely thing about LLMs is that it can handle poorly worded prompts and well worded prompts. On the engineering side, we'll certainly see more rigor and best practices. For your average user? They can keep throwing whatever they like at it.

Exactly. I have been using OpenAI for taking transcriptions and finding keywords/phrases that belong to particular categories. There are existing tools/services that do this – but I would need to learn their API.

With OpenAI, I described it in English, provided sample JSON that I would like, run some tests, adjust and then I am ready.

There was no manual to read, it is in my format, and the language is natural.

And that is what I like about all this -- putting folks with limited technical skills in power.

Post reply on HN