I feel like Chris way, way understates the prior art on C language. It's not just two compilers and some textbooks. That would actually be impressive.
If Internet trained, the training data probably has so much stuff about C compilation in it. Books, step by step articles with code, debugging, StackOverflow answers, mailing lists, blog posts, compiler results... endless. It's one of the most specified and pretrained things in existence for a multi-TB, Internet-trained LLM.
A real test for Claude Code Compiler would be something that had hardly any search results. Then, the few results it had were the language description, some examples, and maybe the existing compiler. Could it output a compiler for that? Can it do ZL (C++ in Scheme), Pony, Mercury, or Cray's Chapel?
Even easier, Lattner should try to have it write a LLVM-compatible compiler for Mojo. Then, lots of multithreaded, SIMD, and GPU implementations of business and ML algorithms in Mojo. That might not only be a good demonstrator but help the business, too.
"Is Google hurting itself in its confusion? Google is the largest scraper in the world. Google's entire business began with a web crawler that visited every publicly accessible page on the internet, copied the content, indexed it, and served it back to users. It did this without distinguishing between copyrighted and non-copyrighted material, and it did this without asking permission. Now Google is in federal court claiming that our scraping is illegal."
The above was my take on the situation, too. If SerpStack is illegal, then Google is illegal.
If SerpStack has to cease to operate or pay damages, nearly everyone with a web server will be collecting a check from Google as they're being liquidated.
1. Generic, mask layers and board to handle what's common across models. Especially memory and interface.
2. Specific layers for the model implementation.
Masks are the most expensive part of ASIC design. So, keeping the custom part small with the rest pre-proven in silicon, even shared across companies, would drop the costs significantly. This is already done in hardware industry in many ways but not model acceleration.
Then, do 8B, 30-40B, 70B, and 405B models in hardware. Make sure they're RLHF-tuned well since changes will be impossible or limited. Prompts will drive most useful functionality. Keep cranking out chips. There's maybe a chance to keep the weights changeable on-chip but it should still be useful if only inputs can change.
The other concept is to use analog, neural networks with the analog layers on older, cheaper nodes. We only have to customize that per model. The rest is pre-built digital with standard interfaces on a modern node. Given the chips would be distributed, one might get away with 28nm for the shared part and develop it eith shuttle runs.
Bruce Schneier had the Friday Squid Blogging posts that he allowed random stuff on. Most of the noise was contained in the Friday threads.
Lobsters had a weekly thread by caius called, "What are you doing this week?" People would post personal projects or experiences in it. On top of interesting tech, that let us pray more specifically for and encourage some in need.
That's pretty awesome what you're doing. Tough that you got to feeling responsible for all of those people. I've struggled with that weight.
The good news is God doesn't hold us responsible for them. He is sovereignly in control of the entire universe, including every life. He made us to worship and reflect Him, and love others as ourselves day by day. We're only judged for what we can actually do, more from the heart than anything.
That's a huge relief. I just help who I can. It's also awesome that, if we believe in Jesus Christ, we can pray fir them to be healed or for Him to raise up entire organizations to give people more medical care. We've seen some healings but usually exceptional, medical progress. (Some suffer or die.) On more care available, your team was one of our answered prayers. :)
You could charge people to fine-tune or customize models. Maybe synthetic data. Maybe rent it on vast.ai but careful about your bandwidth and energy. I'm not sure of the security implications of vast.ai, either.
It can be 10,000 requests a day on static HTML and non-existent, PHP pages. That's on my site. I'd rather them have Christ-centered and helpful content in their pretraining. So, I still let them scrape it for the public good.
It helps to not have images, etc that would drive up bandwidth cost. Serving HTML is just pennies a month with BunnyCDN. If I had heavier content, I might have to block them or restrict it to specific pages once per day. Maybe just block the heavy content, like the images.
Btw, anyone tried just blocking things like images to see if scaping bandwidth dropped to acceptable levels?
https://news.ycombinator.com/item?id=47009024
I feel like Chris way, way understates the prior art on C language. It's not just two compilers and some textbooks. That would actually be impressive.
If Internet trained, the training data probably has so much stuff about C compilation in it. Books, step by step articles with code, debugging, StackOverflow answers, mailing lists, blog posts, compiler results... endless. It's one of the most specified and pretrained things in existence for a multi-TB, Internet-trained LLM.
A real test for Claude Code Compiler would be something that had hardly any search results. Then, the few results it had were the language description, some examples, and maybe the existing compiler. Could it output a compiler for that? Can it do ZL (C++ in Scheme), Pony, Mercury, or Cray's Chapel?
Even easier, Lattner should try to have it write a LLVM-compatible compiler for Mojo. Then, lots of multithreaded, SIMD, and GPU implementations of business and ML algorithms in Mojo. That might not only be a good demonstrator but help the business, too.
reply