Hacker News: fnbr

New comment by fnbr in "Olmo 3: Charting a path through the model flow to lead open-source AI"

fnbr — Fri, 21 Nov 2025 16:52:25 +0000

(I'm a researcher on Olmo.)

There's a bunch of other fully open models, including the [Marin](https://marin.community/) series of models out of Stanford and Nvidia regularly releases fully open models.

New comment by fnbr in "Olmo 3: Charting a path through the model flow to lead open-source AI"

fnbr — Fri, 21 Nov 2025 16:32:18 +0000

MoEs have a lot of technical complexity and aren't well supported in the open source world. We plan to release a MoE soon(ish).

I do think that MoEs are clearly the future. I think we will release more MoEs moving forward once we have the tech in place to do so efficiently. For all use cases except local usage, I think that MoEs are clearly superior to dense models.

New comment by fnbr in "Olmo 3: Charting a path through the model flow to lead open-source AI"

fnbr — Fri, 21 Nov 2025 15:06:25 +0000

(I’m a researcher on the post-training team at Ai2.)

Where did you try this? On the Ai2 playground?

New comment by fnbr in "Olmo 3: Charting a path through the model flow to lead open-source AI"

fnbr — Fri, 21 Nov 2025 15:05:33 +0000

(I’m a researcher on the post-training team at Ai2.)

7B models are mostly useful for local use on consumer GPUs. 32B could be used for a lot of applications. There’s a lot of companies using fine tuned Qwen 3 models that might want to switch to Olmo now that we have released a 32B base model.

The Bitter Lesson

fnbr — Thu, 26 Jun 2025 20:37:53 +0000

Article URL: https://finbarr.ca/the-bitter-lesson/

Comments URL: https://news.ycombinator.com/item?id=44391078

Points: 2

# Comments: 0

New comment by fnbr in "Open guide to equity compensation"

fnbr — Mon, 14 Apr 2025 01:30:36 +0000

Like what? All the examples people have said are where either

1) the company has Nx preferences, for N >1, in which case the company has essentially failed to fundraise or

2) the company sells for less than they raised, which again, is a polite form of failure.

New comment by fnbr in "Open guide to equity compensation"

fnbr — Sun, 13 Apr 2025 22:52:04 +0000

This is why I will never work somewhere with a short post termination exercise period (PTEP). If it’s not at least 5 years, ideally 10, they don’t seriously consider equity something that employees are owed.

New comment by fnbr in "Open guide to equity compensation"

fnbr — Sun, 13 Apr 2025 22:49:30 +0000

Can you explain? In most cases, preferences won’t come into play, assuming you raise at a standard 1x preference and sell for more than you have raised. In that case, owning 0.5% should roughly translate into $5M (modulo dilution).

How to hire ML engineers/researchers

fnbr — Fri, 17 Jan 2025 16:55:39 +0000

Article URL: https://www.artfintel.com/p/how-to-hire-ml-engineersresearchers

Comments URL: https://news.ycombinator.com/item?id=42740218

Points: 1

# Comments: 0

New comment by fnbr in "US Government wants to make it easier for you to click the 'unsubscribe' button"

fnbr — Mon, 12 Aug 2024 13:38:11 +0000

I hate how every company that I place an order with treats that as permission to send a constant drip of marketing emails. I send them straight to spam.

New comment by fnbr in "Character.ai CEO Noam Shazeer Returns to Google"

fnbr — Fri, 02 Aug 2024 20:12:49 +0000

They raised ~$200M total, and $150M at the series A. So if they're just paying back investors, they'd "only" need $500M.

New comment by fnbr in "Character.ai CEO Noam Shazeer Returns to Google"

fnbr — Fri, 02 Aug 2024 18:15:54 +0000

do we know they were profitable? I doubt it, if they pulled this. I think they had high DAU/MAU but low paid users.

this is basically Inflection 2.0.

New comment by fnbr in "Ask HN: Is it possible to make FAANG salaries without working there?"

fnbr — Tue, 23 Jul 2024 13:20:40 +0000

Yes, many AI startups pay more than FAANG for ML talent.

New comment by fnbr in "Stripe closed our account and refuses to migrate our data to another processor"

fnbr — Thu, 30 May 2024 22:15:58 +0000

What's the dating app?

New comment by fnbr in "Codestral: Mistral's Code Model"

fnbr — Wed, 29 May 2024 15:20:27 +0000

The rule of thumb is roughly 44gb, as most models are trained in bf16, and require 16 bits per parameter, so 2 bytes. You need a bit more for activations, so maybe 50GB?

you need enough RAM and HBM (GPU RAM) so it’s a constraint on both.

OpenAI departures: Why can’t former employees talk?

fnbr — Fri, 17 May 2024 18:55:02 +0000

Article URL: https://www.vox.com/future-perfect/2024/5/17/24158478/openai-departures-sam-altman-employees-chatgpt-release

Comments URL: https://news.ycombinator.com/item?id=40393121

Points: 1254

# Comments: 961

New comment by fnbr in "How to Make More Money Renting a GPU Than Nvidia Makes Selling It"

fnbr — Fri, 03 May 2024 19:17:29 +0000

Agreed, that’s what I’m seeing too.

New comment by fnbr in "How to Make More Money Renting a GPU Than Nvidia Makes Selling It"

fnbr — Fri, 03 May 2024 14:27:32 +0000

These prices are insane- you can do much, much better than those prices if you negotiate. If you're paying $7.20 per GPU-hour for an H100, you should find another supplier.

New comment by fnbr in "An open source initiative to share and compare heat pump performance data"

fnbr — Fri, 12 Apr 2024 21:59:13 +0000

Yeah Alberta has very little hydro, while Ontario/Quebec/BC have lots, so our electricity generates a lot more carbon.

New comment by fnbr in "How Hertz’s bet on Teslas went sideways"

fnbr — Thu, 04 Apr 2024 18:09:55 +0000

That's a great reason not to have electric rental cars, but not a great reason to not charge rental cars.