rapatel0

7 ms·

61.
by rapatel0 1y ago
Duckdb can run as a local instance that points to parquet files in a n s3 bucket. So your "auth" can live on the layer that gives permissions to access that bucket.
62.
by rapatel0 1y ago
I think that this is actually the biggest threat to the current "AI bubble." Model efficiency and diffusion of models to open source. It's probably to start hedging bets on Nvidia
63.
by rapatel0 1y ago
If robots are a better partner, then maybe your are not that great a partner...
64.
by rapatel0 1y ago
Edit: my last calculation is wrong. I inverted it 350 Million will be allocated for indirects (650 Million for research)
65.
by rapatel0 1y ago
It's not misinformation. You are repeating a misleading talking point. Here's what happens. - Professor & students get a grant application for 100K. - University charges indirects at a ratio (0.55) - 155K gets transferred from
66.
by rapatel0 1y ago
> I read or heard someplace that at many universities tuition paid by students in the social sciences is effectively subsidizing the STEM fields I'm very skeptical of this claim. In fact up until a recent funding method change from
67.
by rapatel0 1y ago
Indeed this problem could become worse. Dark patterns are darker when you cannot see them at all
68.
by rapatel0 1y ago
I suggest you xpost to Bryan Johnson's Blueprint community. I think might help you get a lot more food tested. That community is probably ICP and also can amplify the message.
69.
by rapatel0 1y ago
Highly recommend Xyce (maintained by Sandia National Labs). It's one of the only Spice variants that is probably good enough to handle Semiconductor circuit modeling.
70.
by rapatel0 1y ago
I lived this previously. The author is missing some important context. Spray-and-Pray Algorithms After AlexNet, dozens of companies rushed into medical imaging. They grabbed whatever data they could find, trained a model, then pushed it thr
71.
by rapatel0 1y ago
The real point of leverage for here is performance/size. Getting traction in the open weights space kinda forces that the models need to innovate on efficiency. This means the open weight models may get leverage that the closed weight
72.
by rapatel0 1y ago
The transformation function in jpeg (DCT) is generally well defined math. While lossy, most of the information is reprocudable. An LLM is layers and layers of non-linear transformations. It's hard to say exactly how information is accu
73.
by rapatel0 1y ago
In my life I've had the following experiences: - In grad school, I averaged 4 hours of sleep (6/7 days per week) and about 8 hours on sunday for about 5 months straight. - In my first startup, I worked 9am to 11pm (had to walk bac
74.
by rapatel0 1y ago
This is brilliant marketing and strategy from Cloudflare. They are pointing out bad actors and selling a service where they can be the private security guards for your website. I think there could be something interesting if they made a cac
75.
by rapatel0 1y ago
I don't think the research is as clear cut as the article suggests. Firstly, the concept was created in the 60s and only became wide spread in the 80s. The study was conducted in the 1975. This has two important implications: - There
76.
by rapatel0 1y ago
As a father trying to get my kindergartner to read and also someone working in ML, it's amazing to me how this mirrors my life experience and ml concepts. When I was a kid, there was a big effort to experiment on our grade using a conc
77.
by rapatel0 1y ago
Memories are about density. I my memory isn't playing tricks with me, a TCAM is about ~300-400F^2 where F is the feature size of the node. On a per bit level, that means that this bit is 4E^10 bigger. Put another way, the TLB in a CPU
78.
by rapatel0 1y ago
Geez if this works. It makes TCAMs free. Ouch found the killer it takes up 0.1 mm^2 in area. That's a show stopper. Hopefully they can scale it down or use it for server infra.
79.
by rapatel0 1y ago
THIS is so true but also not limited to ML. Having been both a publisher and reviewer across multiple engineering, science, and bio-medical disciplines this occurs across academia.
80.
by rapatel0 1y ago
Can you intelligently bake in Claude code router? That would help both with token budget but more importantly multiplexing between different models.
81.
by rapatel0 1y ago
I literally lived this with GDPR. In the beginning every one ran around pretending to understand what it meant. There were a ton of consultants and lawyers that basically made up stuff that barely made sense. They grifted money out of star
82.
by rapatel0 1y ago
We don't know what degree of pressure was or was not exerted on state authorities to compel them to support ICE. Also, I don't think sharing data would be considered enforcing federal law.
83.
by rapatel0 1y ago
My comment shouldn't be read as supporting the activities or really disagreeing with your statement. Was merely indicating that the legal authorities are not as clear cut as the article suggests. To use your examples: 1) enforcement of
84.
by rapatel0 1y ago
The supremacy clause of the constitution asserts that federal law takes precedence over state laws. There are thousands of state laws on the books that are basically rendered null, because a federal law overrides it. One clear example is se
85.
by rapatel0 1y ago
The supremacy clause of the constitution asserts that federal law takes precedence over state laws. There are thousands of state laws on the books that are basically rendered null, because a federal law overrides it. One clear example is se
86.
by rapatel0 1y ago
I respect both the comment and the commenter, but this is a fundamentally speculative statement that is somewhat meaningless. It paraphrases to "it wasn't intentional, but something was intentional, and also unpredictable." I
87.
by rapatel0 1y ago
The key insight here is "too compliant to user requests" What likely happened is that a few people decided to query groq in to generate ragebait traffic to the page/account/etc. Then critical mass happened to make it go
88.
by rapatel0 1y ago
>True randomness is challenging to achieve, because computers are inherently predictable Most modern CPUs now contain a true RNG. They usually use some combination of a metastable latch, or thermal randomness through some kind of analog
89.
by rapatel0 1y ago
This is nonsense. Your credibility’s (and mine) come from the body of work and associated experience. Not a stamp from an org that costs x dollars per year. I would argue that planes are falling out of the sky because management types belie
90.
by rapatel0 1y ago
As someone with a PhD in electrical engineering, I care exactly zero percent about a paid for stamp from an org calling an me engineer or not.

More ›