_aavaa_

8 ms·

121.
by _aavaa_ 4mo ago
Short of outlawing all models, I don’t see a world in which we ever go back to only manual coding.
122.
by _aavaa_ 4mo ago
*why would they warn if it they didn’t believe it. 2 words: regulatory capture. Ban Chinese models because they’re “bad”. (can’t have head to head competition). Bad open weight model (can’t let people escape the subscription model).
123.
by _aavaa_ 4mo ago
Aside from these being assertions rather than arguments, this is from April of 2025 (I.e forever ago).
124.
by _aavaa_ 4mo ago
Which also means they’ll ask a subscription to use a text editor.
125.
by _aavaa_ 4mo ago
The article spends a good deal of time making the point that these rooms are getting more scarce since people can't afford to volunteer their time .
126.
by _aavaa_ 4mo ago
> Anthropic said it identified a campaign by DeepSeek and two other Chinese AI labs to illicitly extract capabilities from its Claude AI platform to improve their own models Oh, won’t someone think of the poor mass copyright infringers.
127.
by _aavaa_ 4mo ago
Oops, you’re right. My brain understood Mac Studio for some reason.
128.
by _aavaa_ 4mo ago
For an fp4 quantization it should fit with room to spare for KVCache
129.
by _aavaa_ 4mo ago
Depends on the task. The dumber the model the more you need to cut up the task into smaller chunks. Also, if you’re using deepseek it’s worth checking out opencode go ($10 for $60 of equivalent api pricing is a good deal).
130.
by _aavaa_ 4mo ago
This is literally a repeat of the whole “China only make low quality cheap stuff” argument. “All they do is copy.” And now, oops they are world leaders in EVs, batteries, solar, drones, just to name a few on the biggest consumer facing th
131.
by _aavaa_ 4mo ago
This is not a great example depending on how you frame this. The solution to the cheating is, as has always been, to have tests conducted in person, on paper without digital technology, under strict supervision.
132.
by _aavaa_ 4mo ago
What are you talking about? AI is currently better at the things it can do than the average random person on the streets.
133.
by _aavaa_ 4mo ago
Because literally any other AI cannot have the same privileges that Siri AI. The DMA allowed actual browser choice, it can also allow real AI choice.
134.
by _aavaa_ 4mo ago
This is a joke right? Like how you can download any browser on iOS as long as they are all running on Apple’s engine.
135.
by _aavaa_ 4mo ago
On the front page right now is the newest announcement from Xiaomi serving large model at over 1,000 tok/s on standard server gpus. Every facet of the field is being pushed on and advanced at the same time.
136.
by _aavaa_ 4mo ago
Because even in a field like software engineering where the output of our work is save in version control, measuring baseline productivity is hard . LoC: people argue it’s not what’s important PRs/day: same as LoC Getting projects don
137.
by _aavaa_ 4mo ago
I wish they explained why mips can’t be used. All there is is a throwaway statement about it.
138.
by _aavaa_ 4mo ago
That’s the exact excuse used by the publisher in the article.
139.
by _aavaa_ 4mo ago
If you don’t trust the CLI version to be properly sandbox d, why would the desktop one be?
140.
by _aavaa_ 4mo ago
But your definition of processed foods has now shifted to being one of hyperpalitability and likelihood of overeating. Those are two completely different categories. The first is, ostensibly, about the processing done to the ingredients to
141.
by _aavaa_ 4mo ago
Sure my point is that what we’re actually talking about is the ingredients themselves, not how they’re processed. Except it’s through this roundabout way; if your worry is sodium benzoate, it doesn’t matter if the food was extruded, deep fr
142.
by _aavaa_ 4mo ago
Where am I missing that from? The linked Wikipedia article explicitly states that it’s not designed for this. How is the food unhealthy? By having lots of fats? Or high salt? Or high sugar? Is it perhaps the ingredients that make it unhealt
143.
by _aavaa_ 4mo ago
I understand the difference. Do you understand that the classification is not based on healthy/unhealthy but based on how much “processing” was done to the food?
144.
by _aavaa_ 4mo ago
From the wikipedia page: > The Nova definition of ultra-processed food does not comment on the nutritional content of food and is not intended to be used for nutrient profiling. > Nutrient profiling: also nutritional profiling, is the
145.
by _aavaa_ 4mo ago
I don't find this bug absurd. It affects linux kernels <5.6 which reached EOL sometime before 2021. > the use of AI is a way to make it easier on himself. I don't see what's wrong with either this or his response in tha
146.
by _aavaa_ 4mo ago
A lot of this water "usage" is actually used by the electricity generation. So it's an unfair comparison to compare to the basic water needs of a person.
147.
by _aavaa_ 4mo ago
First off, the state of the art models are perusing a strategy of doing more with more, but there are lots of small models that can be run locally on high end pcs. Those small models are more capable, are cheaper, and are faster than the
148.
by _aavaa_ 4mo ago
Maybe, maybe not. But his argument here is about cost, not profit. And on the cost front he is objectively wrong, and could easily see this if he cared to look. But he won’t since that would prevent him from making his bombastic content.
149.
by _aavaa_ 4mo ago
It is entirely misplaced based on the evidence so far. Neither of the bugs you link are currently traced back to a commit that used AI.
150.
by _aavaa_ 4mo ago
No, we can be rational and critique its existence. Especially given the current administration’s track record.

More ›