leonickson

13 ms·

1.
by leonickson 1mo ago
no Anthropic involvement, I just used Claude Code heavily while building this and putting that in the README felt more honest than not mentioning it. Now that I think about it may be it shuld be "built with claude code" instead of
2.
by leonickson 1mo ago
It's tunable, --cache-gb N on the CLI. In my sweep the speed barely moved between a 1GB and 6GB cache (43% vs 70% hit rate, same tok/s) because right now the bottleneck is GPU dispatch, not the SSD. so more RAM doesnt buy much yet
3.
by leonickson 1mo ago
agree, prefill is the weak spot right now. it goes through the same per-token path as decode, which is dumb for long prompts. The fix is on the list: during prefill we can batch the expert reads for the whole prompt per layer instead of per
4.
by leonickson 1mo ago
Thanks
5.
Show HN: Run an 80B Qwen in 4.3 GB of RAM on a Mac, and a 35B on an iPhone(github.com)
312 points by leonickson 1mo ago|141 comments
6.
by leonickson 2mo ago
It is free, for higher limits you have to sign up. Thanks. I have updated the terms, and all the community submissions are under MIT.
7.
by leonickson 2mo ago
Thanks! From what I've seen, LLMs confidently generate SwiftUI that looks off from what you want on the first try, then you're spending a couple of rounds getting it to fix it. I am just making it easier. These are all real, compi
8.
Show HN: 1,250 SwiftUI components, and an MCP that writes them into your app(nibware.dev)
10 points by leonickson 2mo ago|6 comments
9.
by leonickson 2mo ago
The extension is live in Firefox - https://addons.mozilla.org/addon/tomesphere/
10.
by leonickson 2mo ago
Yes, DCR is supported (WorkOS exposes the registration endpoint). What error is Aperture throwing? I suspect a metadata-discovery detail.
11.
by leonickson 2mo ago
I really like your work with the gradients.
12.
by leonickson 2mo ago
Thank you so much! I'd genuinely like to hear more, you can email me here - help@tomesphere.com
13.
by leonickson 2mo ago
Hello, 1. Yes, I am working on it, I will soon release it for Firefox and Safari (by this weekend). 2. For now I am just trying to see what people want, may be the business model is more focused on mcp for higher limits for labs, lets see w
14.
by leonickson 2mo ago
I am sorry, it should have taken you to the paper page/article, I will investigate it, but I did the same search and clicked on the first link, it to me to this - https://tomesphere.com/paper/2301.01365 Thanks for
15.
by leonickson 2mo ago
Will check it, it will be really useful. Thanks
16.
by leonickson 2mo ago
It is all open, bulk-downloadable sources. Here are some links. - arXiv: bulk on S3 (https://info.arxiv.org/help/bulk_data_s3.html) - PubMed Central Open Access: AWS Open Data (s3://pmc-oa-opendata). -
17.
by leonickson 2mo ago
The project has four parts, and I think you may have used the search in the navigation bar. That search is for the paper page (detailed information about that specific paper). When you search, you get a list of results, and if you click one
18.
by leonickson 2mo ago
Hi, I love the genre too. Cosmograph is wonderful, I did try it, but because of its license restriction I could not use it for this project. I do agree that beyond an initial "that's cool" this map may not contribute much, &q
19.
by leonickson 2mo ago
Hello, I agree with you, viz are just cool and might not really have a usecase. In this project map is not the product, it is 1 of 4 parts and to be honest the least important. The value is what is under each dot, the enriched page (TLDR, g
20.
Show HN: I mapped 8.5M research papers into an interactive atlas(tomesphere.com)
85 points by leonickson 2mo ago|28 comments
21.
Show HN: ModelSweep - Open-Source Benchmarking for Local LLMs(github.com)
2 points by leonickson 6mo ago|0 comments