bt1a

5 ms·

31.
by bt1a 1y ago
Perhaps it's the finetune of Opus/Sonnet/whatever that is being served to the feds that is the source of the refusal :)
32.
by bt1a 1y ago
Lolz
33.
by bt1a 1y ago
I have been using gpt-5 through the API a bit recently, and I somewhat felt this response behavior, but it's definitely confirming to hear this from another. It's much more willing (vs gpt-4*) to tell me im a stupid piece of shxt
34.
by bt1a 1y ago
I would challenge the utility of this mental model as again they're not simply tracing a "most likely" path unless your sampling methods are trivially greedy. I don't know of a better way to model it, and I promise I&#x2
35.
by bt1a 1y ago
optimizing for the shortest path from idea to prod sounds a tad warped, if i may
36.
by bt1a 1y ago
Do you use an application launcher / configuration manager like Lutris to do this? Or do you mean directly through steam? There's a steam game that I play often that tends to work the most frequently with proton hotfix for reasons
37.
by bt1a 1y ago
i'd wager your benchmark problems require cumbersome arithmetic or are poorly worded / inadequately described. or, you're mislabeling them as basic math and logic (a domain within which LLMs have proven their strengths!) i on
38.
by bt1a 1y ago
giggled like a child through this one
39.
by bt1a 1y ago
these 3090s are mine. hands off!
40.
by bt1a 1y ago
Yet theyre fantastic personal tutors / assistants who can provide a deeply needed 1:1 learning interface for less privileged individuals. I emphasize 'can'. Not saying kids should have them by their side in their current roug
41.
by bt1a 1y ago
Well, they're usually told that they're some unicorn master of * languages, frameworks, skillsets, etc., so can you really fault them? :)
42.
by bt1a 1y ago
Bluffing is a legitimate strategy at times.
43.
by bt1a 1y ago
You hittin them balloons again, mate ?
44.
by bt1a 1y ago
... trust me.. this time it really was a cosmic ray bitflip
45.
by bt1a 2y ago
Heh Can be turned off by the user usually means you must run your own recursive DNS resolver and firewall
46.
by bt1a 2y ago
looks great on mobile, too. it's exceptional at organizing the content's contrast ratios and spacing, my only nitpick is slightly too much color
47.
by bt1a 2y ago
It really is best in slot. Owe it to git, which has a particular synergy with a hallucination-prone but correctable system
48.
by bt1a 2y ago
In a selfish sense, the act of writing can be beneficial to one's inner thinking, regardless of the quality of output.
49.
by bt1a 2y ago
You've beautifully put what swirls vaguely in my mind. They're useful, fallible tools with extraordinary function when operating within known and reasonable tolerances of error
50.
by bt1a 2y ago
How excellent for a quantized 27GB model (the Q6_K_L GGUF quantization type uses 8 bits per weight in the embedding and output layers since they're sensitize to quantization)
51.
by bt1a 2y ago
DeepSeek-R1-Distill-Qwen-32B-Q6_K_L.gguf solved this: In which of the following Incertae sedis families does the letter `a` appear the most number of times? ``` Alphasatellitidae Ampullaviridae Anelloviridae Avsunviroidae
52.
by bt1a 2y ago
and many of these people havent debugged messages more complex than a Python error message. tastelessly jabbing at needing to earn your marks by slamming into segfaults and pushing gdb
53.
by bt1a 2y ago
sudo sysctl iogpu.wired_limit_mb=45000
54.
by bt1a 2y ago
I concur, and I appreciate the pun :)
55.
by bt1a 2y ago
To be fair, nothing comes close to Qwen2.5 atm
56.
by bt1a 2y ago
could it be that the rooster/cockadoodledoo is something performed like clockwork in the morning, so the showing framedata also being tangential to time and clocks ? probably not
57.
by bt1a 2y ago
im curious about the process to find that initial buffer address - does that involve entering a few different strings and searching the memory snapshot for those byte patterns ?
58.
by bt1a 2y ago
What is the importance of needing to frame Python as tiny here? The reminder that programming languages are acceptable for small throwaway scripts is appreciated
59.
by bt1a 2y ago
Yeah, as someone who had access to gpt-4 early in 2023, the endpoint used to take over a minute to respond and the quality of the responses was mindblowing. Simply too expensive to serve at scale, not to mention the silicon constraints that
60.
by bt1a 2y ago
This is beautiful! It is how I look at mindfulness mediation, and I will use some of the clear writing from this piece when trying to convey it to others

More ›