The best Hacker News stories from Show from the past day
Latest posts:
Show HN: San Francisco Compute – 512 H100s at <$2/hr for research and startups
Hey folks! We're Alex and Evan, and we're working on putting together a 512 H100 compute cluster for startups and researchers to train large generative models on.
- it runs at the lowest possible margins (<$2.00/hr per H100)
- designed for bursty training runs, so you can take say 128 H100s for a week
- you don’t need to commit to multiple years of compute or pay for a year upfront<p>Big labs like OpenAI and Deepmind have big clusters that support this kind of bursty allocation for their researchers, but startups so far have had to get very small clusters on very long term contracts, wait months of lead time, and try to keep them busy all the time.<p>Our goal is to make it about 10-20x cheaper to do an AI startup than it is right now. Stable Diffusion only costs about $100k to train -- in theory every YC company could get up to that scale. It's just that no cloud provider in the world will give you $100k of compute for just a couple weeks, so startups have to raise 20x that much to buy a whole year of compute.<p>Once the cluster is online, we're going to be pretty much the only option for startups to do big training runs like that on.
Show HN: San Francisco Compute – 512 H100s at <$2/hr for research and startups
Hey folks! We're Alex and Evan, and we're working on putting together a 512 H100 compute cluster for startups and researchers to train large generative models on.
- it runs at the lowest possible margins (<$2.00/hr per H100)
- designed for bursty training runs, so you can take say 128 H100s for a week
- you don’t need to commit to multiple years of compute or pay for a year upfront<p>Big labs like OpenAI and Deepmind have big clusters that support this kind of bursty allocation for their researchers, but startups so far have had to get very small clusters on very long term contracts, wait months of lead time, and try to keep them busy all the time.<p>Our goal is to make it about 10-20x cheaper to do an AI startup than it is right now. Stable Diffusion only costs about $100k to train -- in theory every YC company could get up to that scale. It's just that no cloud provider in the world will give you $100k of compute for just a couple weeks, so startups have to raise 20x that much to buy a whole year of compute.<p>Once the cluster is online, we're going to be pretty much the only option for startups to do big training runs like that on.
Show HN: San Francisco Compute – 512 H100s at <$2/hr for research and startups
Hey folks! We're Alex and Evan, and we're working on putting together a 512 H100 compute cluster for startups and researchers to train large generative models on.
- it runs at the lowest possible margins (<$2.00/hr per H100)
- designed for bursty training runs, so you can take say 128 H100s for a week
- you don’t need to commit to multiple years of compute or pay for a year upfront<p>Big labs like OpenAI and Deepmind have big clusters that support this kind of bursty allocation for their researchers, but startups so far have had to get very small clusters on very long term contracts, wait months of lead time, and try to keep them busy all the time.<p>Our goal is to make it about 10-20x cheaper to do an AI startup than it is right now. Stable Diffusion only costs about $100k to train -- in theory every YC company could get up to that scale. It's just that no cloud provider in the world will give you $100k of compute for just a couple weeks, so startups have to raise 20x that much to buy a whole year of compute.<p>Once the cluster is online, we're going to be pretty much the only option for startups to do big training runs like that on.
Show HN: LLMFlows – LangChain alternative for explicit and transparent apps
Hi HN! Over the last several weekends, I've been building LLMFlows as an alternative to langchain.<p>There's been a lot of discussion on the shortcomings of langchain in the past few weeks, but when I first tried it in March, I thought there are 3 main problems:
1. Too many abstractions
2. Hidden prompts and opinionated logic in chains which makes it hard to customize
3. Hard to debug<p>This inspired me to try and build a framework that solves these 3 issues, and therefore I started building LLFlows with the "philosophy" of being "simple, explicit, and transparent."<p>A few weekends later, I think I finally managed to reach a state where I feel it's ready to be shared.<p>I would love to hear your feedback! Thank you!
Show HN: ssh-tpm-agent – SSH agent for TPMs
Show HN: ssh-tpm-agent – SSH agent for TPMs
Show HN: Gogit – Just enough Git (in Go) to push itself to GitHub
Show HN: Gogit – Just enough Git (in Go) to push itself to GitHub
Show HN: Gogit – Just enough Git (in Go) to push itself to GitHub
Show HN: Worst programming language written in less than an hour
Unfinished side project inspared by JavaScript
It's just a stupid interpeter for my poor language
Show HN: Worst programming language written in less than an hour
Unfinished side project inspared by JavaScript
It's just a stupid interpeter for my poor language
Show HN: Envoy playground in the browser
Hey HN,<p>We made an Envoy Proxy playground [0] so we could test out our Envoy configs directly in the browser. This is based on Julia's work with Nginx Playround. [1] We forked that repo and added more Envoy to it. [2] Check it out!<p>[0] - Envoy is a popular programmable proxy similar to Nginx or HAProxy that is popular with cloud-native setups: <a href="https://www.envoyproxy.io" rel="nofollow noreferrer">https://www.envoyproxy.io</a><p>[1] - <a href="https://nginx-playground.wizardzines.com" rel="nofollow noreferrer">https://nginx-playground.wizardzines.com</a><p>[2] - <a href="https://github.com/apoxy-dev/envoy-playground">https://github.com/apoxy-dev/envoy-playground</a>
Show HN: I made a little tool to practice country positions on the world map
Hi HN,<p>I built a little tool to help me get better at remembering country locations on the map: <a href="https://learn-worldmap.netlify.app/" rel="nofollow noreferrer">https://learn-worldmap.netlify.app/</a><p>It's made from plain HTML/CSS/JS, and not intended to make any money or anything. Just thought that maybe someone else may enjoy it. Also of course happy about any feedback!<p>Cheerz.
Show HN: I made a little tool to practice country positions on the world map
Hi HN,<p>I built a little tool to help me get better at remembering country locations on the map: <a href="https://learn-worldmap.netlify.app/" rel="nofollow noreferrer">https://learn-worldmap.netlify.app/</a><p>It's made from plain HTML/CSS/JS, and not intended to make any money or anything. Just thought that maybe someone else may enjoy it. Also of course happy about any feedback!<p>Cheerz.
Show HN: I made a little tool to practice country positions on the world map
Hi HN,<p>I built a little tool to help me get better at remembering country locations on the map: <a href="https://learn-worldmap.netlify.app/" rel="nofollow noreferrer">https://learn-worldmap.netlify.app/</a><p>It's made from plain HTML/CSS/JS, and not intended to make any money or anything. Just thought that maybe someone else may enjoy it. Also of course happy about any feedback!<p>Cheerz.
Show HN: Chie – a cross-platform, native, and extensible desktop client for LLMs
I'm submitting this before going to bed, so by the time it hits front page (if it does at all) I'll likely be asleep, so I won't be able to answer questions in time.
Show HN: Chie – a cross-platform, native, and extensible desktop client for LLMs
I'm submitting this before going to bed, so by the time it hits front page (if it does at all) I'll likely be asleep, so I won't be able to answer questions in time.
Show HN: Chie – a cross-platform, native, and extensible desktop client for LLMs
I'm submitting this before going to bed, so by the time it hits front page (if it does at all) I'll likely be asleep, so I won't be able to answer questions in time.
Show HN: Selecta – Tune Your Own Spotify Recommendation Algorithm
I posted here a while back with this app I'd been working on. In short:<p>It allows you to talk to Spotify's recommendations API to specify musical features you'd like recommendations similar to.<p>It's loads of fun to mess around with, and I've given it an overhaul to include many more dimensions ito. musical features, as well as the ability to log in with Spotify, and save things you find to a playlist. You don't need a subscription to Spotify to use the app either.<p>Thanks to @Mockapapella for the tagline when I posted the first version: <a href="https://news.ycombinator.com/item?id=35907215">https://news.ycombinator.com/item?id=35907215</a>
Show HN: Selecta – Tune Your Own Spotify Recommendation Algorithm
I posted here a while back with this app I'd been working on. In short:<p>It allows you to talk to Spotify's recommendations API to specify musical features you'd like recommendations similar to.<p>It's loads of fun to mess around with, and I've given it an overhaul to include many more dimensions ito. musical features, as well as the ability to log in with Spotify, and save things you find to a playlist. You don't need a subscription to Spotify to use the app either.<p>Thanks to @Mockapapella for the tagline when I posted the first version: <a href="https://news.ycombinator.com/item?id=35907215">https://news.ycombinator.com/item?id=35907215</a>