czk

6 ms·

61.
by czk 2y ago
> Wiz has raised a total of $1.9 billion from a combination of venture capital funds and private investors > Wiz agreed to acquire Tel Aviv-based Raftt, a cloud-based developer collaboration platform, for $50 million in December 2023.
62.
by czk 2y ago
This reminds me of Renaissance Technologies (Jim Simons, PhD in math from uc berkeley). He started the company in 1982 after he left academia for finance and leveraged quant models for training, they mostly hire PhD mathematicians, physicis
63.
by czk 2y ago
The only time I've seen extensive citing from an AI is when I use Deep Research with the ChatGPT Pro plan. Otherwise, they are mostly few and far between. Perplexity was one of the early services I used that seemed to try to make an ef
64.
by czk 2y ago
Can't help but comment that the song Honey Bee by Tom Petty is an absolute banger as well.
65.
by czk 2y ago
> “Many of these algorithmic secrets, there are $100 million secrets that are a few lines of code" I wonder how many western labs have not hesitated to implement techniques revealed openly by DeepSeek. DeepSeek has open-sourced argu
66.
by czk 2y ago
I have had great success with Claude Code by asking it to do TDD (something I personally don't practice). The agentic loop seems to benefit from writing the tests first and then hands-off code iteration until they pass. When working wi
67.
by czk 2y ago
China’s population is packed into dense urban areas and high-rise buildings, unlike the sprawling suburbs of the US... Could this mean fewer distinct “dots of light” on the map, even if economic activity is high? It looks like measuring ove
68.
by czk 2y ago
Yeah you’re right— sounds like accounts made prior to Feb 2018 were lower than 100,100. Yikes! Updated my post.
69.
by czk 2y ago
Horrible, and waiting until Christmas week to disclose it while weak master passwords (no min length reqs) and a laughable PBKDF2-HMAC-SHA256 with 5,000 iterations (this was set to 100,100 for new accounts after Feb 2018) let the attackers
70.
by czk 2y ago
Never underestimate the power of institutional momentum! cough IBM AS400
71.
by czk 2y ago
SemiAnalysis has a nice write-up on MI300X vs H100/H200 and concludes that the CUDA moat is still very real: https://semianalysis.com/2024/12/22/mi300x-vs-h100-vs-h200-b... "As fast as AMD tries to
72.
by czk 2y ago
could this complaint be generalized to the software quality of anything that's been built upon for many years? as the churn in the workforce happens you lose nuance and expertise and systems become more and more complex to maintain and
73.
by czk 2y ago
its mostly a bunch of stuff i whipped together myself, but the short of it is i have a docker container that acts as a "vpn broker", establishing multiple vpn connections and then exposing access to them over socks proxies, and in
74.
by czk 2y ago
meanwhile deepseek has been releasing absolute fire open-source all week long...
75.
by czk 2y ago
This exactly! Apologies for not being more clear.
76.
by czk 2y ago
I’m kind of addicted to tab containers right now and I have them set up in a way where I can proxy specific containers out different socks proxies that go through different VPN tunnels. Niche I know but it’s keeping me on FF.
77.
by czk 2y ago
Is there a “brave” browser that uses Firefox?
78.
by czk 2y ago
https://archive.is/2025.02.26-000501/https://www.nytimes.com...
79.
by czk 2y ago
I found this very insightful, and quite a good summary and overview of what you can expect from an LLM https://thebullshitmachines.com/
80.
by czk 2y ago
Wow, I never knew they used to offer DDT-laced wallpaper. https://thesocietypages.org/socimages/files/2008/02/DDTwallp...
81.
by czk 2y ago
Dedupe seemed more interesting when storage was expensive, but nowadays it feels like the overhead you get from running dedupe, in most cases, is priced-in. At least with software like CommVault for backups, dedupe requires beefy hardware a
82.
by czk 2y ago
I pay for a number from voip.ms and use sms forwarding. Its very cheap and it works on telegram as well which seemed fairly strict at detecting most voips.
83.
by czk 2y ago
At least you are providing them with valuable training data, then. Maybe in a future model!
84.
by czk 2y ago
Would love to know just how overfit a lot of them are on these benchmarks
85.
by czk 2y ago
can't wait for cloudflare to implement this to further disrupt my browsing freedoms and waste cpu cycles
86.
by czk 2y ago
is it just me or does calling these distilled models 'DeepSeek R1' seem to be a gross misrepresentation of what they actually are people think they can run these tiny models distilled from deepseek r1 and are actually running deep
87.
by czk 2y ago
Got a chuckle from that recent commit message. I feel you
88.
by czk 2y ago
thats far from the case here i really have no skin in the game...
89.
by czk 2y ago
im very naive here but does anyone trust these benchmarks? do they mean anything to you? they seem far too easy to game and it doesn't feel like its an accurate way to really tell how these models compare to one another. seems like ben
90.
by czk 2y ago
the fact that openai hides the reasoning tokens from us to begin with shows that what they are doing behind the scenes isnt all that impressive, and likely easily cloned (r1) would be nice if they made them visible now

More ›