Hacker News: eachro

New comment by eachro in "FunctionGemma 270M Model"

eachro — Thu, 18 Dec 2025 20:43:19 +0000

Do you think this would be appropriate for a command line tool that hits various apis as the function calls? Ex: "what's the weather in SF tomorrow?" Or "daily price change of apple, Tesla stock for past week"? (Let's assume I have documented the apis thoroughly somewhere that the model has access to or fine tuned it on this data)

New comment by eachro in "Integer Programming (1977) [pdf]"

eachro — Mon, 08 Sep 2025 23:38:26 +0000

Does anyone know what the state of the art industry solvers do for these problems? I had dabbled a bit in ml approaches to combinatorial optimization with great interest a few years back, but I don't think any of these rl based methods ended up being used in production.

New comment by eachro in "Amazon has mostly sat out the AI talent war"

eachro — Tue, 02 Sep 2025 02:48:38 +0000

Didn't Amazon aquihire Adept Labs?

New comment by eachro in "Gemma 3 270M re-implemented in pure PyTorch for local tinkering"

eachro — Wed, 20 Aug 2025 17:18:37 +0000

If you wanted to train it from scratch, how long would it take on a reasonable GPU setup?

New comment by eachro in "Fairness is what the powerful 'can get away with' study shows"

eachro — Fri, 15 Aug 2025 14:40:14 +0000

I'm reminded of the nixon quote: "When the president does it, that means it's not illegal."

New comment by eachro in "NYC's office-to-residential conversions could create 17,000 new homes"

eachro — Thu, 17 Jul 2025 18:58:05 +0000

What would it take to make NYC more like Tokyo where you have consumer/retail level things on the not-ground floor level.

New comment by eachro in "Smollm3: Smol, multilingual, long-context reasoner LLM"

eachro — Tue, 08 Jul 2025 18:17:38 +0000

From what I've heard, the llama3 models are fairly easy to fine-tune (please correct me if I'm wrong or if there are more amenable models here). How easy is it to finetune smollm3? I know a lot of the MoE LLMs have been quite fickle in this regard.

New comment by eachro in "Everything around LLMs is still magical and wishful thinking"

eachro — Sat, 05 Jul 2025 07:06:02 +0000

"And 50% of the time they work 50% of the time."

I think this is still an incredible outcome given how many dice rolls you can take in parallel with multiple claude/o3/gemini attempts at a problem with slightly different prompts. Granted, each rollout does not come for free given the babysitting you need to do but the cost is much lower than going down the path yourself/having junior colleagues make the attempt.

New comment by eachro in "Use keyword-only arguments in Python dataclasses"

eachro — Mon, 30 Jun 2025 12:37:43 +0000

Yeah haha I got autocorrected

New comment by eachro in "Use keyword-only arguments in Python dataclasses"

eachro — Mon, 30 Jun 2025 12:16:20 +0000

Is there a reason to use data classes over pedantic base models anymore?

New comment by eachro in "Building an AI server on a budget"

eachro — Mon, 09 Jun 2025 22:11:33 +0000

A lot of people are saying 12gb is too small to do anything interesting with. What's the most useful thing people __have__ gotten to work?

New comment by eachro in "Mathematical Foundations of Reinforcement Learning"

eachro — Tue, 11 Mar 2025 06:07:57 +0000

During the openai gym era of RL, one of the great selling pts was that RL was very approachable for a new comer as the gym environments were small and tractable that a hobbyist could learn a little bit of RL, try it out on cartpole and see how it'd perform. Are there similarly tractable RL tasks/learning environments with LLMs? From the outside, my impression is that you need some insane GPU access to even start to mess around with these models. Is there something one can do on a normal MacBook air for instance in this LLM x RL domain?

New comment by eachro in "Introduction to Stochastic Calculus"

eachro — Mon, 24 Feb 2025 19:17:20 +0000

For those in quant finance, how much of this is useful in your day to day?

New comment by eachro in "FFmpeg School of Assembly Language"

eachro — Sat, 22 Feb 2025 18:24:45 +0000

This looks great! Is there going to be exercises or a project based component as well?

New comment by eachro in "Waymo to test its autonomous driving technology in over 10 new cities"

eachro — Wed, 29 Jan 2025 20:55:02 +0000

Thats a great choice. High trust city, super high density.

New comment by eachro in "Best Pens for 2025"

eachro — Sun, 12 Jan 2025 21:41:35 +0000

Do the rankings ever change much year over year? For instance, with intro fountain pens, it will always be things like the platinum preppy, pilot metro, kaweco sport.

New comment by eachro in "A Gentle Introduction to Graph Neural Networks (2021)"

eachro — Fri, 20 Dec 2024 18:46:55 +0000

Is there consensus about whether gnn architectures are better than transformer based ones at this point? I am aware that transformers can be viewed as a gnn too.

New comment by eachro in "Ask HN: Recommendation for a SWE looking to get up to speed with latest on AI"

eachro — Wed, 27 Nov 2024 15:05:03 +0000

Reproduce nanogpt.

Then find a small dataset and see if you can start getting close to some of the reported benchmark numbers with similar architectures.

New comment by eachro in "The Truth About Linear Regression (2015)"

eachro — Tue, 30 Jul 2024 16:59:34 +0000

I'd love to see linear regression taught by say a quant researcher from Citadel. How do these guys use it? What do they particularly care about? Any theoretical results that meaningfully change the way they view problems? And so on.

New comment by eachro in "Using SIMD for Parallel Processing in Rust"

eachro — Mon, 01 Jul 2024 17:57:51 +0000

This is cool that simd primitives exist in the std lib of rust. I've wanted wanted to mess around a bit more with simd in python but I don't think that native support exists. Or your have to go down to C/C++ bindings to actually mess around with it (last I checked at least, please correct me if I'm wrong).