p0u4a

7 ms·

1.
by p0u4a 2mo ago
> if software knows in advance which data chunk (expert) it'll need for the next token, it can load that in parallel with computing current token You could actually use the model's MTP head to make a ~decent prediction on what
2.
by p0u4a 2mo ago
more of stuff like this on the web please
3.
by p0u4a 4mo ago
I have no idea what's going on but it looks really cool
4.
Antikythera Mechanism (oldest known analogue computer)(historyofinformation.com)
3 points by p0u4a 4mo ago|0 comments
5.
Cloudflare's Project Think(blog.cloudflare.com)
8 points by p0u4a 5mo ago|3 comments
6.
Constructing an LLM Computer(percepta.ai)
3 points by p0u4a 5mo ago|0 comments
7.
by p0u4a 6mo ago
How much does training such a model cost?
8.
Show HN: OpenChat – Syncing conversations across AI providers(github.com)
1 points by p0u4a 6mo ago|0 comments
9.
You can now mount HuggingFace Storage Buckets as local filesystems(github.com)
1 points by p0u4a 6mo ago|0 comments
10.
Deerflow(deerflow.tech)
1 points by p0u4a 6mo ago|0 comments
11.
by p0u4a 6mo ago
Sadly they don't teach Show Don't Tell in the lectures
12.
by p0u4a 6mo ago
Aaaand we’re back
13.
by p0u4a 6mo ago
Edit: I was hosting this on Vercel free tier, and since this post made it to the front page, I've hit Vercel's quota limit. I will work on getting around this.
14.
by p0u4a 6mo ago
I was on Vercel free tier and hit the quota for blob storage. Will probably make it open source so people can self host
15.
by p0u4a 6mo ago
Hmm, yeah I suppose you're right. I'm currently using the algolia api https://hn.algolia.com/ Will look into tightening this up
16.
by p0u4a 6mo ago
yeah that would be awesome, will land it tomorrow
17.
by p0u4a 6mo ago
I've been thinking about this for a while too. Probably what I will do is create embeddings of the frontpage results and then let users pick 3-5 topics that they would like to see. Then summarize the top 10 for those topics.
18.
by p0u4a 6mo ago
Great suggestion. Will add this tomorrow
19.
Show HN: Hackerbrief – Top posts on Hacker News summarized daily(hackerbrief.vercel.app)
82 points by p0u4a 6mo ago|46 comments
20.
In space, no one can hear you kernel panic (2020)(increment.com)
88 points by p0u4a 6mo ago|23 comments
21.
by p0u4a 6mo ago
We're entering the area of stochastic software, everybody's a gambler
22.
by p0u4a 6mo ago
Linear
23.
by p0u4a 6mo ago
not on my bingo sheet that's for sure. The whole verification thing is also so laughably flawed. There needs to be a way for agents to prove they are not being prompted or steered by a human while on the site, which is a non-trivial pr
24.
Mechanical Sympathy(mechanical-sympathy.blogspot.com)
2 points by p0u4a 7mo ago|0 comments
25.
by p0u4a 7mo ago
Someone once told me, agentic coding may lead to something akin to a "software engineering union" forming, where a set of guidelines control code quality. Namely, at least one of writing, testing, and reviewing of code must be don
26.
by p0u4a 7mo ago
For desktop specifically, native is just not worth it anymore. For example setting up permissions for every OS is a pain in native, but you get that for free on the web. You can basically do everything native could offer now, even notificat
27.
by p0u4a 7mo ago
Pretty close to what you describe: https://github.com/fastmachinelearning/hls4ml
28.
by p0u4a 7mo ago
Here’s the repo if anyone’s interested https://github.com/P0u4a/yorha
29.
Recreated the Nier Automata UI in React(yorha-design.vercel.app)
4 points by p0u4a 7mo ago|1 comments
30.
by p0u4a 7mo ago
Flutter keeping Dart alive as always

More ›