Hacker News: jonesn11

New comment by jonesn11 in "Project Glasswing: Securing critical software for the AI era"

jonesn11 — Thu, 09 Apr 2026 17:23:22 +0000

Spell doom.. frfr

Claude Opus 4.5, and why evaluating new LLMs is increasingly difficult

jonesn11 — Thu, 27 Nov 2025 09:04:06 +0000

Article URL: https://simonwillison.net/2025/Nov/24/claude-opus/

Comments URL: https://news.ycombinator.com/item?id=46067294

Points: 6

# Comments: 1

New comment by jonesn11 in "NASA has a list of 10 rules for software development"

jonesn11 — Tue, 18 Feb 2025 00:30:41 +0000

Good thing to give to LLMs as well.

New comment by jonesn11 in "Ingesting PDFs and why Gemini 2.0 changes everything"

jonesn11 — Thu, 13 Feb 2025 20:21:19 +0000

OCR makes sense, but it is another asking for a summary. It is not there yet, gave a lot of incorrect details.

New comment by jonesn11 in "OnlyFans models are using AI impersonators to keep up with their DMs"

jonesn11 — Thu, 19 Dec 2024 00:14:04 +0000

Yeah this is whole operation is weird and I knew it

New comment by jonesn11 in "Procrastination and the fear of not being good enough"

jonesn11 — Thu, 14 Nov 2024 22:52:30 +0000

My personal opinion is that the procrastination the guy is experiencing is due to an underlying mental health challenge

New comment by jonesn11 in "I Don't Have Spotify"

jonesn11 — Thu, 14 Nov 2024 01:36:42 +0000

Spotify literally plays the same songs over and over...and creates features no one wants

New comment by jonesn11 in "Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku"

jonesn11 — Wed, 23 Oct 2024 14:26:00 +0000

How does one get access to it without using the API??

New comment by jonesn11 in "Reproducing GPT-2 in llm.c"

jonesn11 — Wed, 29 May 2024 04:59:51 +0000

Like the FAQ, you correctly anticipated my questions.

New comment by jonesn11 in "Lawsuit against Meta asks if Facebook users have right to control their feeds"

jonesn11 — Thu, 02 May 2024 01:25:32 +0000

We need this lawsuit against X, formerly known as Twitter. That's where it's really bad.

New comment by jonesn11 in "Biden signs TikTok bill into law, starting clock for ByteDance to divest"

jonesn11 — Fri, 26 Apr 2024 04:26:39 +0000

Though, the saddest thing is that it seems like the U.S citizens, (i.e ANY of tiktoks 160 MILLION US users) have absolutely no say in the operations,a yet we actually interact with the app not these old people in Congress. The fact that Biden so swiftly signed the bill too makes me frustrated as I want to vote for him, but he keeps doing or okaying things that are counter to my values.

New comment by jonesn11 in "Biden signs TikTok bill into law, starting clock for ByteDance to divest"

jonesn11 — Fri, 26 Apr 2024 04:24:23 +0000

The algorithm's novelty and recommendation accuracy is so far beyond what other competitors like Google, Snapchat, and Meta have that this seems like a coordinated effort by the private sector to push forth their mediocre products and centralize social media service which I absolutely DETEST. The saddest thing is that it seems like the U.S citizens, (i.e ANY of tiktoks 160 MILLION US users) have absolutely no say in the operations, yet we actually interact with the app not these old people in Congress. The fact that Biden so swiftly signed the bill too makes me frustrated as I want to vote for him, but he keeps doing or okaying things that are counter to my values.

New comment by jonesn11 in "Daniel Kahneman has died"

jonesn11 — Thu, 28 Mar 2024 22:06:39 +0000

The best psychologist of all time has died. RIP

New comment by jonesn11 in "Daniel Kahneman has died"

jonesn11 — Thu, 28 Mar 2024 05:53:41 +0000

NO NO NO NO NO !

New comment by jonesn11 in "23andMe says user data stolen in credential stuffing attack"

jonesn11 — Mon, 09 Oct 2023 18:29:38 +0000

This is why no matter how many interesting ads or tiktoks, I will never do these genetic testing kits. I wouldn't be surprised military is working on CRISPR like infections that target your specific DNA when sprayed in the air.

New comment by jonesn11 in "OpenAI Codex"

jonesn11 — Thu, 12 Aug 2021 01:25:43 +0000

How did it mess up the "make every fifth line bold" prompt?

Also, to follow up on the original comment, AI demos are nice, but being a student of history there are still fundamental challenges with these systems. My skepticism is in how much prompting is really required and how can it understand higher level semantics like code refactoring, reproducible examples, large scale design patterns etc.

This synthesis of sequential symbolic processes and probabilistic neural generation is really exciting though. When the amount of human code edits and tweaking for complex programs goes down from hours to seconds then that's when I'll be impressed and scared.

New comment by jonesn11 in "GitHub Copilot"

jonesn11 — Wed, 30 Jun 2021 13:45:26 +0000

This is an interesting business model. One problem I foresee though is that frameworks can get outdated in a few years, and that might affect the text generation abilities.

New comment by jonesn11 in "Can Transformers Crack the Coding Interview?"

jonesn11 — Tue, 25 May 2021 02:42:19 +0000

Measuring Coding Challenge Competence With APPS Abstract: While programming is one of the most broadly applicable skills in modern society, modern machine learning models still cannot code solutions to basic problems. It can be difficult to accurately assess code generation performance, and there has been surprisingly little work on evaluating code generation in a way that is both flexible and rigorous. To meet this challenge, we introduce APPS, a benchmark for code generation. Unlike prior work in more restricted settings, our benchmark measures the ability of models to take an arbitrary natural language specification and generate Python code fulfilling this specification. Similar to how companies assess candidate software developers, we then evaluate models by checking their generated code on test cases. Our benchmark includes 10,000 problems, which range from having simple one-line solutions to being substantial algorithmic challenges. We fine-tune large language models on both GitHub and our training set, and we find that the prevalence of syntax errors is decreasing exponentially. Recent models such as GPT-Neo can pass approximately 15% of the test cases of introductory problems, so we find that machine learning models are beginning to learn how to code. As the social significance of automatic code generation increases over the coming years, our benchmark can provide an important measure for tracking advancements.

(Source: Dan Hendrycks et al.)

Can Transformers Crack the Coding Interview?

jonesn11 — Tue, 25 May 2021 02:42:19 +0000

Article URL: https://arxiv.org/abs/2105.09938

Comments URL: https://news.ycombinator.com/item?id=27272086

Points: 1

# Comments: 1

New comment by jonesn11 in "Psilocybin 'promising' for depression"

jonesn11 — Sun, 18 Apr 2021 16:20:14 +0000

The confidence intervals don't seem to imply that Psilocybin is "promising" given that for a subset of patients, there were worse effects.