acchow

10 ms·

91.
by acchow 1y ago
I believe gp is saying the MCP’s “tool/list” endpoint should return dynamic, but OpenAPI-format, content. Not that the list of tools and their behavior should be static (which would be much less capable)
92.
by acchow 1y ago
Each unit effectively owns a percentage of the overall land under the building
93.
by acchow 1y ago
What about $/sqft? Or $/acre of actual land?
94.
by acchow 1y ago
“Plural of anecdote is data” is meant to be tongue-in-cheek. Actual data is sampled randomly. Anecdotes very much are not.
95.
by acchow 1y ago
Correct. ZGC has no way to escape from the virtualization by the kernel (assuming your hardware and kernel uses an MMU)
96.
by acchow 2y ago
> Now we seem to have found an AI worthy of that title and instead start moving the goal posts on "consciousness" The goalposts already differentiated between "totally human-like" vs "actually conscious" See
97.
by acchow 2y ago
IIRC, conversion to date happens after editing the cell value.
98.
by acchow 2y ago
A good reason to define it as an invalid program then and fail at compile time?
99.
by acchow 2y ago
> Q4_K_M has ~0 quality degradation Llama 3.3 already shows a degradation from Q5 to Q4. As compression improves over the years, the effects of even Q5 quantization will begin to appear
100.
by acchow 2y ago
This is only for the small model. The medium model is still at 1M (like Gemini 2.5) Even if we could get the mid models to 10M, that's still a medium-sized repo at best. Repos size growth will also accelerate as LLMs generate more code
101.
by acchow 2y ago
Nvidia GPUs can natively operate in FP8, FP6, FP4, etc so naturally they have reduced memory requirements when running quantized. As for CPUs, Intel can only go down to FP16, so you’ll be doing some “unpacking”. But hopefully that is “on th
102.
by acchow 2y ago
> The concept is simple: Take a picture of the food you are about to consume, and let the app log calories and macros for you. > The result is an app that the creators say is 90% accurate, which appears to be good enough for many diet
103.
by acchow 2y ago
How do you compose the cheapest models to create a software engineer?
104.
by acchow 2y ago
Temporal is for a static graph with idempotent nodes. Powerful LLM workflows don’t fit this model.
105.
by acchow 2y ago
Why not? “~= 0” just means we likely will be unable to measure any effect. Not that there literally is no effect. Us being unable to do science on something means we cannot scientifically state that it does not exist. Not that we can cat
106.
by acchow 2y ago
The abstractions are fictional. The system wasn’t “designed” with separation. We just pretend they are there for tractability.
107.
by acchow 2y ago
Waymo is significantly more expensive than UberX nowadays. People are happily willing to pay for the better experience (and tourists probably the novelty)
108.
by acchow 2y ago
The running joke is that nobody in NYC knows their neighbors. In a sense, you no longer need to since you now have thousands of people within about a dozen-suburban-house's distance away.
109.
by acchow 2y ago
Keep in mind, Oracle was designed to run with 128KB of RAM (no swapping). So it was really tens of thousands of lines, not millions.
110.
by acchow 2y ago
Curiously, the things I search most often are not located in files: calendar, photo content/location, email, ChatGPT history, Spotify library, iMessage/whatsapp history, contacts, notes, Amazon order history
111.
by acchow 2y ago
Do you mean FPGAs? Because fixed silicon inevitably has fixed "baked in" choices.
112.
by acchow 2y ago
> Wait just so i understand it, if a single human creates an AI model and trains it, and then prompts it to create an image, is that considered "human intervention" and does that make that human the author of that image? I gues
113.
by acchow 2y ago
How would it be aware of everything else I have and do online? Calendar, emails, YouTube history, current audiobook, etc?
114.
by acchow 2y ago
> Replace the monkey with a 2nd human, and it's obvious that "the guy" does not earn the copyright, it goes to the person who took the photo. If there was no person, then there is no copyright. If I set up an entire scene
115.
by acchow 2y ago
> EPA range estimates being inaccurate is a real problem. They do not, and are not designed to, give actual expected range. It’s meant to be an “average” of “mixed” driving. Also, EPA ranges expect mostly constant speed and driving withi
116.
by acchow 2y ago
If the gradient is steep enough then you would not be recouping most of the energy going back downhill because your brakes will be burning off the energy. Uphill and downhills is a combination where the estimated range might be thrown off e
117.
by acchow 2y ago
Inventory along the distribution chain will take quite some time to draw down before people are actually drinking less coffee. As for caffeine, there are cheaper alternatives to coffee if you just want the caffeine.
118.
by acchow 2y ago
We're in the early days, like back when Uber offered $2.50 rides anywhere in San Francisco. They burned through tons of cash for years but had $9.8bn net income last year.
119.
by acchow 2y ago
> since are no more dimensions to expand into. Quantum computing is next, right?
120.
by acchow 2y ago
History of function calls: - goto/jmp to instr - vtable lookup - hash and lookup in a dictionary - run a Large Language Model

More ›