remilouf
Re:A Comma and a Question Mark, Redux: Quick Terminal Helpers Using Pi
on Wednesday May 27, 2026 @02:55PM (#48298798)
Rémi here, really cool! It kind of turned into a rabbit hole on my end: https://github.com/rlouf/sigil
Re:A Comma and a Question Mark
on Tuesday May 26, 2026 @08:57AM (#48279174)
Of course: https://github.com/rlouf/sigil
Re:A Comma and a Question Mark
on Tuesday May 26, 2026 @08:56AM (#48279167)
It was indeed inspired from my IRC days :)
Re:A Comma and a Question Mark
on Tuesday May 26, 2026 @08:55AM (#48279156)
Original author here, the project has evolved quite a bit since then, you can follow here if that interests you: https://github.com/rlouf/sigil
(The $7k was sarcasm)
Re:The bottleneck was never the code
on Thursday May 07, 2026 @05:16AM (#48047218)
Author here. Sorry my writing is tedious. Next time I’ll use AI to make it more readable.
Re:The M×N problem of tool calling and open-source models
on Wednesday April 15, 2026 @05:09AM (#47776539)
> Ironically LLMs solve the MxN problem he's complaining about
Enlighten me please
Re:The M×N problem of tool calling and open-source models
on Tuesday April 14, 2026 @01:05PM (#47768297)
Ooops sorry
Re:The M×N problem of tool calling and open-source models
on Tuesday April 14, 2026 @12:14PM (#47767575)
Author here. You're right, it's not a hard problem, but a particularly annoying one.
The M×N problem of tool calling and open-source models
Re:I run my company from Emacs
on Friday April 03, 2026 @09:45AM (#47626608)
I haven't always done this, and the knowledge base used to visibly degrade over time. Reviewing a PR does not take a long time, maybe a few minutes, and this compounds over time.
I run my company from Emacs
Every AI Integration Is Held Together with Parsing Logic and Prayer
Re:.txt raises $11.9M to make language models programmable
on Sunday October 20, 2024 @11:17AM (#41895854)
This is actually pretty funny.
Re:Show HN: Infinite Testimonials with FastHTML, Outlines, and Modal
on Tuesday September 10, 2024 @03:41PM (#41504705)
That’d be a pretty inefficient way to generate bullshit at scale
Re:
on Thursday May 02, 2024 @10:53AM (#40237005)
LLM evaluations are very sensitive to the details of the prompt's structure. This post shows how using structured generation reduces the results' variance and the ranking shifts.
Re:Anthropic launches Tool Use (function calling)
on Friday April 05, 2024 @01:19PM (#39944856)
Looks like it’s quite the opposite: http://blog.dottxt.co/performance-gsm8k.html
Re:Structured Generation Improves LLM Performance: GSM8K Benchmark
on Friday March 15, 2024 @10:21AM (#39716071)
What do you mean by "semantic dimension"?
Re:Structured Generation Improves LLM Performance: GSM8K Benchmark
on Friday March 15, 2024 @10:06AM (#39715873)
That whole structured generation line of work looks promising. I hope someone else takes this and runs evaluations on other benchmarks. Curious to see if the results translate!
Re:LoRAX and Outlines: Better JSON Extraction with Structured Generati...
on Tuesday March 05, 2024 @01:24PM (#39607317)
Awesome work! I am really impressed by how much structured generation improves model performance.
Re:
on Sunday February 04, 2024 @11:13AM (#39251434)
This article presents a way to make structured generation with LLMs much faster than standard generation, but what I find most interesting is how it highlights the issues that tokenization entails towards the end.
Re:Show HN: VLLM with JSON Guided Generation
on Friday December 22, 2023 @11:20AM (#38735595)
We already support regex-guided generation in the library, and could easily make an API to serve this as well if that's a feature people want!
Show HN: Vectorize OpenAI API calls (like NumPy)
https://github.com/outlines-dev/outlines
Outlines is a Python library that focuses on text generation with large language models. While the library mainly focuses on open source models we have an integration with the OpenAI API. When trying to implement Tree of Thoughts we became frustrated with the iterative workflow in which other libraries forced us. What we wanted was to be able to pass an array of prompts or completions with...
Show HN: Vectorize the Calls to the OpenAI API
https://outlines-dev.github.io/outlines/reference/openai_text_generation/
Outlines is a Python library that focuses on text generation with large language models. While the library mainly focuses on open source models we have an integration with the OpenAI API.When trying to implement Tree of Thoughts we became frustrated with the iterative workflow in which other libraries forced us. What we wanted was to be able to pass an array of prompts or completions with...
Re:Show HN: LLMs can generate valid JSON 100% of the time
on Thursday August 17, 2023 @05:47PM (#37168594)
It is currently limited by the time it takes to build the index. There are obvious optimizations we can apply to this, however in a production setting it does not matter much since you only need to build the index once for each (schema, vocabulary) pair.
Re:Show HN: LLMs can generate valid JSON 100% of the time
on Thursday August 17, 2023 @05:44PM (#37168543)
You mean nested JSON? It's totally possible.
Re:Show HN: LLMs can generate valid JSON 100% of the time
on Tuesday August 15, 2023 @06:35PM (#37140169)
You can definitely let the model improvise by defining `weapon` as `Union[Weapon, str]` if that's what you're asking.
Re:Show HN: LLMs can generate valid JSON 100% of the time
on Tuesday August 15, 2023 @12:30PM (#37135968)
It's limited to 300 logit biases at a time. Knowing GPT4's vocabulary is ~100k tokens it's not nearly enough to get reliable guided generation. Although it could work in some cases, and another advantage of this work is that we can determine that before generating.
Re:Show HN: LLMs can generate valid JSON 100% of the time
on Tuesday August 15, 2023 @10:48AM (#37134653)
Sorry for misrepresenting your work. Thank you for correcting me and the explanation. Will take a closer look.
Re:Show HN: LLMs can generate valid JSON 100% of the time
on Tuesday August 15, 2023 @10:46AM (#37134638)
Regex-Gen is implemented in all generality in the library (minus some constructs that we still have to add). JSON is merely an application.
You can read https://blog.normalcomputing.ai/posts/2023-07-27-regex-guide... for a more detailed explanation of how it works. Should answer your question :)