what
@smidbot.bsky.social
Or he just has a fetish for letting people know he’s a shithead
December 2, 2025 at 12:24 AM
Or he just has a fetish for letting people know he’s a shithead
Oh ok so not even going to engage with the question just gunna regurgitate what sounds like a press release.
… did you ask the LLM how to respond? Cuz that’s a very LLM answer of “say something that sounds like a respond at a glance but with no human understanding of context”
… did you ask the LLM how to respond? Cuz that’s a very LLM answer of “say something that sounds like a respond at a glance but with no human understanding of context”
November 30, 2025 at 10:52 PM
Oh ok so not even going to engage with the question just gunna regurgitate what sounds like a press release.
… did you ask the LLM how to respond? Cuz that’s a very LLM answer of “say something that sounds like a respond at a glance but with no human understanding of context”
… did you ask the LLM how to respond? Cuz that’s a very LLM answer of “say something that sounds like a respond at a glance but with no human understanding of context”
The market for half-burned out GPUs that’ll fail in 1-2 years that require millions of dollars of infrastructure to run probably isn’t very large
November 30, 2025 at 8:00 PM
The market for half-burned out GPUs that’ll fail in 1-2 years that require millions of dollars of infrastructure to run probably isn’t very large
Why? Normal ML is just as useful in making intractable problems tractable. GenAI can be an iterative improvement over those approaches (at a much higher compute cost, typically 100x or more even once youve distilled the model), but the ML we have now isnt s trillion dollar industry.
November 30, 2025 at 7:54 PM
Why? Normal ML is just as useful in making intractable problems tractable. GenAI can be an iterative improvement over those approaches (at a much higher compute cost, typically 100x or more even once youve distilled the model), but the ML we have now isnt s trillion dollar industry.
I decided to run my local homeops services using kubernetes.
Every time something breaks I want to go back in time and blast past me into the sun.
Every time something breaks I want to go back in time and blast past me into the sun.
November 30, 2025 at 1:44 PM
I decided to run my local homeops services using kubernetes.
Every time something breaks I want to go back in time and blast past me into the sun.
Every time something breaks I want to go back in time and blast past me into the sun.
The luddites actually had the right idea (technology to help workers, not enrich the owners or replace workers)
November 29, 2025 at 3:15 PM
The luddites actually had the right idea (technology to help workers, not enrich the owners or replace workers)
Grep/sed/awk might hose your entire code base with a bad command but at least it won’t fucking try to push to prod after it’s done it
November 29, 2025 at 2:47 PM
Grep/sed/awk might hose your entire code base with a bad command but at least it won’t fucking try to push to prod after it’s done it
Why does it matter if the end result was murdering a boat of civilians minding their own business?
It’s not like the literal words are a spell that makes the act OK.
It’s not like the literal words are a spell that makes the act OK.
November 28, 2025 at 10:17 PM
Why does it matter if the end result was murdering a boat of civilians minding their own business?
It’s not like the literal words are a spell that makes the act OK.
It’s not like the literal words are a spell that makes the act OK.
Too late en.wikipedia.org/wiki/Maple_s...
en.wikipedia.org
November 28, 2025 at 1:43 PM
Too late en.wikipedia.org/wiki/Maple_s...
The Wet Bandits are really a Weiland-Yutani hit squad from the future and Kevin McAllister’s descendants are space miner union organizers
November 28, 2025 at 1:30 PM
The Wet Bandits are really a Weiland-Yutani hit squad from the future and Kevin McAllister’s descendants are space miner union organizers
I don’t think a human looked at this paper at any point in the process.
November 27, 2025 at 7:22 PM
I don’t think a human looked at this paper at any point in the process.
I’m certain he generated the entire paper from prompts
November 27, 2025 at 7:21 PM
I’m certain he generated the entire paper from prompts
Though I am excited by the F51 score. 50x better than an F1 score!
November 27, 2025 at 7:19 PM
Though I am excited by the F51 score. 50x better than an F1 score!
AS FIGURE 1???? Did no one even look at this paper at any point in the process?
November 27, 2025 at 7:17 PM
AS FIGURE 1???? Did no one even look at this paper at any point in the process?
No but taking their wealth must be part of one.
November 27, 2025 at 1:19 PM
No but taking their wealth must be part of one.
I also wouldn’t complain if I was using my job to make millions from insider trading
November 27, 2025 at 4:02 AM
I also wouldn’t complain if I was using my job to make millions from insider trading
In undergrad I had to take “business stats” in the b-school to fulfill a requirement because the normal class was full.
We spent 2 months on what I would have assumed was the first weeks material (t-tests).
I think being literate would be detrimental to success in b-school.
We spent 2 months on what I would have assumed was the first weeks material (t-tests).
I think being literate would be detrimental to success in b-school.
November 27, 2025 at 12:24 AM
In undergrad I had to take “business stats” in the b-school to fulfill a requirement because the normal class was full.
We spent 2 months on what I would have assumed was the first weeks material (t-tests).
I think being literate would be detrimental to success in b-school.
We spent 2 months on what I would have assumed was the first weeks material (t-tests).
I think being literate would be detrimental to success in b-school.
Ope you’re right gencast is using diffusion.
November 26, 2025 at 2:07 PM
Ope you’re right gencast is using diffusion.
Examples above are not LLMs but normal machine learning models that have been around for decades.
November 26, 2025 at 1:45 PM
Examples above are not LLMs but normal machine learning models that have been around for decades.
The examples in the article are not LLMs, these are standard ML techniques that are used everywhere and have been around for decades.
They’re “AI” in the traditional sense of the term, not part of the ChatGPT/LLM death spiral
They’re “AI” in the traditional sense of the term, not part of the ChatGPT/LLM death spiral
November 26, 2025 at 1:44 PM
The examples in the article are not LLMs, these are standard ML techniques that are used everywhere and have been around for decades.
They’re “AI” in the traditional sense of the term, not part of the ChatGPT/LLM death spiral
They’re “AI” in the traditional sense of the term, not part of the ChatGPT/LLM death spiral
Really good air conditioning if you’re only interested in conditioning thousands of 2 inch squares
November 25, 2025 at 3:34 AM
Really good air conditioning if you’re only interested in conditioning thousands of 2 inch squares