Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: What are you working on? (February 2025)

news.ycombinator.com

Post reply on HN
Print view View original

541–550 of 922 posts

Re: Ask HN: What are you working on? (February 2025)

#541

Post by fasteddie31003 »

I'm scratching my own itch with https://buildersqrcodes.com to help convey new construction details to job site workers. I'm building my own house now and I was surprised how many details are not in the plans that are critical to build a house. I think this is a common issue. I already have 10's of paying customers using it on their build too.

Re: Ask HN: What are you working on? (February 2025)

#543

Post by maattdd »

I've been updating my HN bot (watch comments for keywords and post to Slack/Discord) written in Crystal to use raw SQL instead of unmaintained ORM.

Turns out the whole app needs only ~ 10 SQL requests, and it's way funier to write modern SQL than fighting the ORM.

The new code looks like this :

    db_message = Model.save_new_message!(@conf.db, DbButler::Hn, item, DbState::Processing)
Than I have a Model module with all the interactions with the DB

    def self.save_new_message!(db, butler, external_id, state) : Message
        sql = {{ read_file "./db/save_new_message.sql" }}
        db.query_one(sql, butler.to_s.downcase, external_id, state.to_s.downcase, as: Message)
   end
(thanks to Crystal ability to read a file at compile time - I can write raw SQL in a file with syntax highlithing and maybe typesafe if I connect the DB to the editor)

The land page is not ready, but the bot has been working for me for months https://newsbutler.xyz/

Re: Ask HN: What are you working on? (February 2025)

#546

Post by AJRF »

niek_pas wrote: post #394
AJRF wrote: post #127

I recently made a little tool for people interested in running local LLMs to figure out if their hardware is able to run an LLM in GPU memory. https://canirunthisllm.com/

Feature request: I would like to know if I can run _any_ LLms on my machine, and if so, which.

I've now had multiple people ask for this - I will work on adding a new tab for this feature as it is a little different than what the site was originally intended to do.

Generally speaking models seem to be bucketed by param count (3b, 7b, 8b, 14b, 34b, 70b) so for a given VRAM bucket you will end up being able to run 1000's of models - so is it valuable to show 1000s of models?

My bet is "No" - and what really is valuable is like the top 50 trending models on HuggingFace that would fit in your VRAM bucket. So I will try build that.

Would love your thoughts on that though - does that sound like a good idea?

Re: Ask HN: What are you working on? (February 2025)

#547

Post by bratao »

I´m super excited, sleepless for a couple of days already. I´m trying to use all tricks possible to improve a Sequence Labeling using Conditional Random Fields. I need to NER billion of documents, and need to be fast. CRFSuite is a workhorse, and a baseline very hard to beat with speed and precision. But with o3 I´m created a frank-stain with many tricks such as CRF with variable order, feature interactions, bidirectional, jointly learning with word embeddings. The precision is already over than CRFSuite. And I believe that would be better than many other solutions such as bi-lstm-crf. Definitely much faster.

Now i´m trying to port to Cython to make as fast as possible. Here o3 is almost useless, but I´m progressing.

Re: Ask HN: What are you working on? (February 2025)

#548

Post by AJRF »

kristopolous wrote: post #532
AJRF wrote: post #127

I recently made a little tool for people interested in running local LLMs to figure out if their hardware is able to run an LLM in GPU memory. https://canirunthisllm.com/

This looks closed source, am I correct?

Not so much purposefully closed source more that I don't want to make it complex by splitting out the data the app uses from the code (co-ordination problem when it comes to deploying that I don't want to deal with for a project of this size).

When it comes to "how to do the math" this repo was my starting point: https://github.com/Raskoll2/LLMcalc

Re: Ask HN: What are you working on? (February 2025)

#550

Post by WoodenChair »

I’m selling refurbished and upgraded Mac mini G4s as the ultimate machines for running software for the classic Mac OS over at https://os9.shop

Over at macos9lives.com a group of hackers figured out a way to get Mac OS 9 running on these late model G4s that previously never supported it. That combined with an SSD upgrade makes them close to the fastest machines that can run Mac OS 9.

I’ve taken advantage of this hack, now having sold about 80 -90 machines. But I’ve hit a wall with finding ways to advertise it. eBay has been okay. I tried Reddit Ads on the vintage Apple subreddit and they were so so—probably lost money doing it but got the word out. Google Search ads have surprisingly been ineffective. I’ve posted on various vintage Mac forums but they don’t allow formal advertising (otherwise I would buy it). I probably will try Facebook ads next. Open to other advertising ideas!

Post reply on HN