stevenae
Re:Pro Max 5x quota exhausted in 1.5 hours despite moderate usage
on Sunday April 12, 2026 @02:23PM (#47742720)
The quantitative ux research team at Google was created for exactly this problem: a service which became popular before the right metrics existed, meaning metrics need to be derived first, then optimized. We would observe users (irl), read their logs, then generate experiments to improve the behavior as measured by logs, and return to see if the experiment improves irl experiences. There were not many of us and we are around :)
Re:Show HN: The Hessian of tall-skinny networks is easy to invert
on Thursday January 15, 2026 @05:53PM (#46640575)
This helped me, coming from an ml background: https://randomrealizations.com/posts/xgboost-explained/
Re:Show HN: OfferGridAI – side-by-side comparison of real estate offer...
on Friday January 02, 2026 @11:22AM (#46466360)
Others mentioned county data. If you can get that, you can build something like I did for DC -- https://colab.research.google.com/drive/1Kep_9j_PN_SxX85PYHE...
Re:Zero-Shot Forecasting: Our Search for a Time-Series Foundation Model
on Saturday June 14, 2025 @10:08AM (#44276534)
My reading of this situation is that MAPE would do the opposite. Means are skewed towards outliers.
Re:Zero-Shot Forecasting: Our Search for a Time-Series Foundation Model
on Friday June 13, 2025 @11:23AM (#44269372)
Thanks for the reply! I am outside the forecasting sphere.
RMSLE gives proportional error (so, scale-invariant) without MAPE's systematic under-prediction bias. It does require all-positive values, for the logarithm step.
Re:Zero-Shot Forecasting: Our Search for a Time-Series Foundation Model
on Friday June 13, 2025 @08:58AM (#44268104)
To clarify, you'd prefer rmsle?
Re:Deep learning gets the glory, deep fact checking gets ignored
on Wednesday June 04, 2025 @11:27AM (#44181789)
For this and sibling -- yes. Essentially, using the output of any model as an input to another model is transfer learning.
Re:Deep learning gets the glory, deep fact checking gets ignored
on Wednesday June 04, 2025 @08:24AM (#44179986)
> Lately, I just steal embeddings from big models and slap a dumb classifier on top. Works better, runs faster, less drama.
You may know this but many don't -- this is broadly known as "transfer learning".
Re:Former Supreme Court Justice David Souter Dies at 85
on Friday May 09, 2025 @04:02PM (#43940413)
Thank you!
Re:Former Supreme Court Justice David Souter Dies at 85
on Friday May 09, 2025 @01:22PM (#43939153)
How accurate is his claim that Augustus became emperor through (my paraphrasing) democratic means and promises to fix real problems for Romans?
Re:Silicon Valley, Halt and Catch Fire, and How Microserfdom Ate the W...
on Wednesday April 02, 2025 @09:08AM (#43556260)
This still strikes me as escapism.
Re:Any insider takes on Yann LeCun's push against current architectures?
on Wednesday March 12, 2025 @06:25PM (#43348309)
Re:Ask HN: Advice for tolerating the "corporate" aspect of software jobs?
on Wednesday March 12, 2025 @08:28AM (#43342505)
There was a saying at Google, I code for free, they pay me for XYZ (literally everything else).
Re:Cameras of 1930s Era
on Thursday February 06, 2025 @08:47AM (#42962242)
I guess my quibble is with the percentage, then. A good, cheap, plentiful camera belies the idea that only the top 0.1% of cameras were good.
Re:Cameras of 1930s Era
on Thursday February 06, 2025 @07:32AM (#42961746)
Disagree with the first piece about only using the top 0.1%. I grew up (through my 20's) shooting on a Pentax K1000, cheap workhorse of a camera, and I preferred its ergonomics to top-end mirrorless cameras I use today.
Re:Ask HN: Who wants to be hired? (January 2025)
on Saturday January 04, 2025 @05:42PM (#42598214)
Location: Washington DC USA
Remote: Hybrid or Remote
Willing to relocate: NYC
Technologies: r, python, sql
Résumé/CV: https://www.linkedin.com/in/steven-ellis-4b140533
Data scientist looking to continue product-focused work!
Re:A “meta-optics” camera that is the size of a single grain of salt
on Friday November 22, 2024 @08:27AM (#42213666)
Pro cameras do not do this to any degree.
Edit: by default.
Re:Is Python That Slow?
on Friday November 22, 2024 @07:35AM (#42213326)
This is not true. R shines for classical stats and ML. If you are doing deep learning, you need Python.
Re:Python, C++ inspired language that transpiles to C and can be embed...
on Thursday November 14, 2024 @02:25PM (#42140114)
From the guidelines [1]: Please don't sneer, including at the rest of the community.
Re:Guys Hacked AirPods to Give Their Grandmas Hearing Aids
on Thursday November 14, 2024 @01:59PM (#42139840)
Save you some scrolling (link directly to comment):
Re:Debugging Compiled Code for R with Positron
on Friday November 01, 2024 @02:15PM (#42019924)
Fellow R / Polars user here. How (in what applications) do you use DuckDB?
Re:New Mac Mini with M4
on Tuesday October 29, 2024 @12:08PM (#41985708)
Airplay? Lunadisplay?
Re:Cookware company includes artificial customer prompts in website
on Saturday August 10, 2024 @11:16AM (#41210115)
Very much seems to be the case
Re:Ask HN: Who wants to be hired? (August 2024)
on Thursday August 01, 2024 @06:12PM (#41134173)
Location: Washington, DC
Remote: Yes, EST-PST
Willing to relocate: Limited
Technologies: Python/R, SQL, MapReduce, Causal Inference, Experimentation, Tabular ML
Résumé/CV: https://www.linkedin.com/in/steven-ellis-4b140533/
Email: On LinkedIn
I am a data scientist and machine learning engineer with several years of experience across startups, corporations, and the federal government. I first learned to program as a child with Logo, then as a teenager with C++ and Java, settling now on Python after stints with R and Javascript.
Re:Ask HN: Boss thinks I'm dumb, now what?
on Monday June 17, 2024 @03:59PM (#40710352)
Have dealt with this a couple of times. I see you're feeling some kind of reaction. I recommend talking it out with someone (professional therapist or otherwise). Often times a frustrating relationship has a clear path to amelioration, but it takes "seeing with fresh eyes".
Re:First randomized trial of Ozempic for alcoholism shows big drops in...
on Monday June 17, 2024 @02:31PM (#40709270)
Not suggesting you, or sibling comment, use this, but posting for awareness: https://www.404media.co/people-in-the-philippines-are-callin...
Re:First randomized trial of Ozempic for alcoholism shows big drops in...
on Monday June 17, 2024 @02:29PM (#40709249)
I'm not disagreeing with that, just stating how they define the term.
Re:First randomized trial of Ozempic for alcoholism shows big drops in...
on Monday June 17, 2024 @02:27PM (#40709225)
I haven't seen this reported. Do you have a source?
Re:First randomized trial of Ozempic for alcoholism shows big drops in...
on Monday June 17, 2024 @02:22PM (#40709154)
Exclusion Criteria:
...
- Current engagement in alcohol treatments, or currently engaged in intentional efforts to quit alcohol use
...