HNHacker News
TopNewBestAskShowJobs

newfocogi

1,059 karma · joined June 5, 2018

submissionscomments
newfocogi··on Koheesio: Nike's Python-based framework to build advanced data-pipelines
I'd like to understand what data engineering inside Nike is actually like. I'm curious because I have relevant experience on my LinkedIn profile, and I get reached out to almost weekly from third party recruiters trying to fill really low paying contract data engineering and ML jobs with Nike. These roles seem to be targeting people with professional experience in the US but pay roughly a 3rd of what I would consider the going rate. There's another top level comment here that this tool might make sense "in a shop with a lot of inexperienced devs", which would confirm my anecdata. Maybe the roles are actually scams, who knows :shrug:
newfocogi··on VC Says Half of Google Staff Do 'No Real Work'
Communication costs scale super-linearly with organization size, as does the cost of mistakes. I think a twitter warrior saying some portion of a massive company's engineers 'Do No Real Work' is just a lazy way to get headlines. Throwing shade at the issue isn't novel, but solving it is really, really hard.
newfocogi··on Chinchilla Scaling: A replication attempt
Key claims:

"We have found three potential issues with Hoffmann et al.’s estimates of the Chinchilla scaling law that rely on Approach 3: 1. Their estimated model fits the reconstructed data very poorly. These conclusions hold even when accounting for potential noise in data reconstruction and excluding outlier models. 2. The confidence are implausibly tight given the number of data points. Obtaining confidence intervals that tight would require many hundreds of thousands of observations, while they likely had only ∼400. 3. Their estimated model implies a scaling policy that is inconsistent with their other approach"

Data point most people are probably looking for: "We find a range consistent with the 20 tokens per parameter rule of thumb. Indeed, our point estimates imply that 25.6 tokens per parameters is optimal."

newfocogi··on Tesla Q4 2023 Earnings Report [pdf]
May be a typo but I’m seeing 10,000 next gen vehicles per week, which is an order of magnitude less than you stated. Otherwise, thanks for the great summary!
newfocogi··on Online auto dealer Vroom announces shutdown
“We announced January 22nd, 2024, that Vroom has halted all purchases and sales of used vehicles. We are discontinuing Vroom's e-commerce operations and winding down our used vehicle dealership business.”
newfocogi··on Show HN: SapientML – Generative AutoML for Tabular Data
Are there benchmarks or third party evaluations that you can share that support your claim? I haven’t used all these offerings so my experience is anecdotal, but I haven’t seen commercial offerings outperform AutoGluon, at least for Tabular data.
newfocogi··on Show HN: SapientML – Generative AutoML for Tabular Data
I’m not sure how the authors can claim to be comparing their approach against “state of the art” AutoML and not include AutoGluon, FLAML or H2O. This independent benchmarking paper[0] is what the AutoML field points to as establishing SOTA, and the libraries compared against in the SapientML paper are middle of the pack at best.

[0] https://arxiv.org/abs/2207.12560

newfocogi··on Chinchilla’s death
While the article makes good observations, this would appear to be a major oversight by leading research labs if they could have just kept the gas pedal down on simpler models for longer and gotten better performance. This is HackerNews – can we get someone from OpenAI, DeepMind, or MetaAI to respond and justify why cutting off the smaller models at a lower total compute budget is justified?
newfocogi··on Google testing ChatGPT-like chatbot 'Apprentice Bard' with employees
While I agree with this sentiment, I think we should be careful assuming that our jobs as knowledge workers are much more than “retaining and regurgitating information”. Even the emotional intelligence and organizational strategy portions of what we do may boil down to this as well.
newfocogi··on US economy returned to growth last quarter, expanding 2.6%
Could you explain why this is good for the renter with a ton of cash?
newfocogi··on TabPFN: Transformer Solves Small Tabular Classification in a Second
It's exciting to see a novel approach to applying NNets to Tabular ML. Definitely need to call out the 1000 row limitation. It will be interesting to see if this approach stands the test of time. Other algorithms (SAINT cough, cough) made big claims, but AFAIK no one actually uses them. It's still a "XGBoost is all you need" world in tabular ML (unless you've discovered AutoGluon).
newfocogi··on Sound waves used to separate microplastics from laundry wastewater
I worked on a project using a similar concept for concentrating micro-algae in the water solution it is grown in. The previous method of extracting micro-algae was filtration or centrifuges. Filters get full too quickly and need regular attention, and centrifuges take too much energy since the density of the particles were very close to the density of water and didn't spin out easily.

We tried using standing acoustic waves to organize and "conclomerate" the micro-algae in the flow and siphon off water quite similarly to the image in the article. It "sort of" worked but only when the flow rate was really, really low because these forces are totally dominated by drag forces and even the smallest levels of turbulence remixes the solution.

What else could this be used for? Think of something that doesn't play nicely with filters or centrifuges and doesn't need to be processed at really high flow rates. Applications in the medical industry come to mind.

newfocogi··on Sound waves used to separate microplastics from laundry wastewater
My guess would be that they choose the frequency based on the speed of sound in the liquid such that the nodes in the standing wave occur at a designed location in the chamber where the separation occurs.
newfocogi··on Sound waves used to separate microplastics from laundry wastewater
Resonant waves in a liquid can cause particles of different sizes to move toward the nodes or anti-nodes of a standing wave. A secondary effect is the particles can conglomerate and fall out depending on density.
newfocogi··on The Surprising Number of Programmers Who Can’t Program
As a software developer, I spend much, much more time reading code, trying to understand code, looking for bugs in code, and tweaking existing code, than I do writing totally new software applications. Although I agree it is surprising that many people can't reason through fizzbuzz, it is very possible to make valuable contribution to a code base in a language that you don't have the syntax for the "main" function memorized.
newfocogi··on Solitude and Leadership (2010)
Artificial General Intelligence
newfocogi··on I rode all the e-scooters, and most of them are awful except two
Does anyone know who the manufacturer of the “Wheels” scooter/bike is? That design appeals to me much more than the standing scooters.
newfocogi··on On Startup Capital Efficiency
Why is every dollar from a customer worth $5 from an investor, as opposed to $2 or $10. I understand the idea behind revenue dollars being more valuable than investment dollars, but I question how the ratio is constant across startups rather than heavily dependent on other factors like industry, stage, round, etc.
newfocogi··on Nordic Valley’s brilliant, zero-budget solution to trail/lift status
In my mind the piece that makes this so appropriate is “The backend is the frontend”. Does anyone know of other platforms/services where this approach is possible?
newfocogi··on People Who Eat the Same Meal Every Day
I am guessing gp meant "surfing" as in the internet, not waves.
newfocogi··on Google Edge TPU Devices
The new Tegra Xavier kits costs ~$1300 and were first released to developers a few months ago
newfocogi··on John C. Bogle Has Died
Bogle is the reason a lot of bad investing decisions are passively avoided.
newfocogi··on Simple Encrypted Arithmetic Library goes open source
I am not an encryption expert, but to the casual techy follower, homomorphic encryption seems like it could be huge for the union of machine learning and digital privacy. Are there any experts on HN who could comment on how mature this technology is? Or does anyone know of any interesting projects that use this? I know Numerai is using something of this sort, but it is hard for me to tell how legitimate it is.
newfocogi··on 'Talent Wants Transit': Companies Near Transportation Gaining the Upper Hand
I don't necessarily think "talent" wants transit. Something that matters to every job seeker is what "home" is going to look like, and being close to transit opens up more types of "home" to choose from. Some talent wants the pace of downtown while others want 4 bedroom houses with yards. Being near transit does a better job of providing both of those to candidates, thereby increasing the talent pool's size.
newfocogi··on Amazon Plans to Split HQ2 Evenly Between Two Cities
Humorous and insightful, the best kind of commentary.

For those who aren't familiar with the reference:

https://en.wikipedia.org/wiki/Judgment_of_Solomon

newfocogi··on Ask HN: What are some intellectually stimulating video games?
Kerbal Space Program
newfocogi··on Is Saudi money becoming radioactive?
Obligatory reference to Betteridge's Law of Headlines... But seriously, the idea that you vote with your dollars becomes even more compelling when you have billions to throw around. Unfortunately when you cast your vote and what you end up voting for don't always happen in the expected order.
newfocogi··on Stop writing lambda expressions in Python
The author is entitled to his perspective, but I would equate this article to JS articles titled "A framework author's advice to avoid frameworks". Its a trendy article because it takes something that a lot of people use and describes its weaknesses. Fun to read and argue with, but not a mainstream idea.
newfocogi··on Ask HN: Best robust laptop for everyday software development
In the future, your computer will go out for you.
newfocogi··on Apple Special Event [video]
It was interesting to hear apple state that one of their goals for building products was to make them last longer in order to minimize waste. People often accuse hardware manufacturers, including apple, of intentional obsolescence. I am interested to see if they actually change anything about their process, software and hardware, to put any weight behind this claim, or if it is just marketing doublespeak.
← PreviousPage 3 of 5Next →