Hacker News: mlepath

A Boy Who Cried AGI

mlepath — Mon, 27 Jan 2025 14:16:27 +0000

Article URL: https://medium.com/@ibreznik/the-boy-who-cried-agi-8b37b1382197

Comments URL: https://news.ycombinator.com/item?id=42841348

Points: 5

# Comments: 1

New comment by mlepath in "Show HN: I made an open-source laptop from scratch"

mlepath — Thu, 23 Jan 2025 15:55:46 +0000

This is an awesome project! Thanks for taking time to document this. What's next on your plate? How do we follow you?

New comment by mlepath in "Enterprises in for a shock when they realize power and cooling demands of AI"

mlepath — Wed, 15 Jan 2025 16:35:35 +0000

All companies that are doing something with the current iteration of "AI" are underwater. Sam Altman says that the pro tier of ChatGPT is losing money, Adobe is losing tons on firefly, ... This is pretty typical for silicon valley though, we always burn investor money to corner the market and then tech usually catches up. Most enterprises don't need to be first adopters.

New comment by mlepath in "Don't use cosine similarity carelessly"

mlepath — Wed, 15 Jan 2025 04:58:20 +0000

In ML everything is a tradeoff. The article strongly suggests using dot product similarity and it's a great metric in some situations, but dot product similarity has some issues too: - not normalized (unlike cosine simularity) - heavily favors large vectors - unbounded output - ...

Basically, do not carelessly use any similarity metric.

New comment by mlepath in "Tabby: Self-hosted AI coding assistant"

mlepath — Mon, 13 Jan 2025 03:31:02 +0000

Awesome project! I love the idea of not sending my data to a big company and trust their TOS.

The effectiveness of coding assistant is directly proportional to context length and the open models you can run on your computer are usually much smaller. Would love to see something more quantified around the usefulness on more complex codebases.

New comment by mlepath in "AI founders will learn the bitter lesson"

mlepath — Sun, 12 Jan 2025 17:00:38 +0000

The author appears to be confused about the difference between research and production. In research more generic approaches typically win because they get resourced much better (plus the ever-growing compute and data have helped, but there is no guarantee that these will continue).

On production side of "AI" (I don't love the term being thrown around this loosely as true AI should include planning, etc not just inference) the only question is how well do you solve the one problem that's in front of you. In most business usecases today that problem is narrow.

LLMs drive a minuscule (but growing) amount of value today. Recommender systems drive huge amount of value. Recommender systems are very specialized.

New comment by mlepath in "Phi 4 available on Ollama"

mlepath — Sun, 12 Jan 2025 05:32:21 +0000

> This reinforce the idea that there is no real strategic advantage in owning a model

For these models probably no. But for proprietary things that are mission critical and purpose-built (think Adobe Creative Suite) the calculus is very different.

MS, Google, Amazon all win from infra for open source models. I have no idea what game Meta is playing

New comment by mlepath in "What the US Pornhub 'ban' is really about"

mlepath — Sun, 12 Jan 2025 05:27:59 +0000

Wait, I thought every pornsite already had age verification, that modal that pops up and says "are you sure you are over 18?"

New comment by mlepath in "Why the Singularity Is Impossible – Or Why It's Just a Bad Word in General"

mlepath — Sat, 11 Jan 2025 14:54:27 +0000

As ML practitioner I am not a big fan of the concept of singularity and wholly agree that IF it happens it will be more like sunrise than a light switch, but...

I think thinking about it IS useful. Problems like value alignment are really difficult. No one alive knows what to do about value alignment and even if it takes us 500 years to reach real singularity it may take that long to solve value alignment problem

New comment by mlepath in "Meta's memo to employees rolling back DEI programs"

mlepath — Fri, 10 Jan 2025 23:06:10 +0000

I worked at Meta, I am not sure the hiring was ever "diverse".

DEI always seemed like an activity they did for show. This changes nothing honestly.

New comment by mlepath in "Show HN: TabPFN v2 – A SOTA foundation model for small tabular data"

mlepath — Fri, 10 Jan 2025 14:20:23 +0000

Great work!

Do you see any artifacts from having trained on synthetic data? Is there a natural benchmark dataset (real tables in the wild)?

In my experience synthetic data can only take you so far, it has all the quirk the dataset creator can think of but the real value is usually in patterns they cannot. Vision took a huge leap forward with ImageNet dataset release

New comment by mlepath in "Show HN: HipScript – Run CUDA in the browser with WebAssembly and WebGPU"

mlepath — Wed, 08 Jan 2025 04:51:25 +0000

This is really awesome. Tested it on both Nvidia and Mac GPUs.

Interested to know how debugging in a real application would work since WASM is pretty hard to debug and GPU code is pretty hard to debug. I assume WASM GPU is ... very difficult to debug.

New comment by mlepath in "Time-Series Anomaly Detection: A Decade Review"

mlepath — Tue, 07 Jan 2025 03:44:07 +0000

The process-centric taxonomy in this paper is one of the most structured frameworks I’ve seen for anomaly detection methods. It breaks down approaches into distance-based, density-based, and prediction-based categories. In practice (been doing time series analysis professionally for 8+ years), I’ve found that prediction-based methods (e.g., reconstruction errors in autoencoders) are fantastic for semi-supervised use cases but fall short for streaming data.

Nvidia Launches AI Foundation Models for RTX AI PCs

mlepath — Tue, 07 Jan 2025 03:37:53 +0000

Article URL: https://nvidianews.nvidia.com/news/nvidia-launches-ai-foundation-models-for-rtx-ai-pcs

Comments URL: https://news.ycombinator.com/item?id=42618910

Points: 3

# Comments: 0

New comment by mlepath in "How I program with LLMs"

mlepath — Tue, 07 Jan 2025 01:58:13 +0000

The first rule of programming with LLMs is don't use them for anything you don't know how to do. If you can look at the solution and immediately know what's wrong with it, they are a time saver otherwise...

I find chat for search is really helpful (as the article states)

New comment by mlepath in "'Precocious' early-career scientists with high citation counts proliferate"

mlepath — Sun, 05 Jan 2025 05:18:57 +0000

At University I once talked to a former Soviet professor about publications and he mentioned how they proliferated. It was uncommon to have a lot of publications until recently and now everyone has a really high count.

I think we came to any metric that is used for rewards loses its value as a metric. That includes citations, sadly.

New comment by mlepath in "My 25-year adventure in AI and ML"

mlepath — Thu, 02 Jan 2025 04:21:57 +0000

I have had a somewhat similar journey but 14 years instead of 25 and I always wonder how it would be different today.

We were lucky enough to grow up with the industry and progressively learn more complexity. The kids out of school today are faced with decades worth of complexity on day one on the job.

New comment by mlepath in "How AI is unlocking ancient texts"

mlepath — Wed, 01 Jan 2025 23:36:10 +0000

This is a great application of various domains of ML. This reminds me of Vesuvius Challenge. This kid of thing is accessible to beginners too since the data by definition are pretty limitted.

New comment by mlepath in "Terence Tao: One of my papers got declined today"

mlepath — Wed, 01 Jan 2025 23:32:35 +0000

Hello, fellow Metamate ;)

New comment by mlepath in "Nvidia bets on robotics to drive future growth"

mlepath — Wed, 01 Jan 2025 16:01:13 +0000

I have worked with Jetson Orin platform, and honestly Nvidia has something that is really easy to work with there. The Jetsons are basically a full GPU (plus some stuff) at very low power. If I were tasked with building a robot it would likely be the first place I look.