@gianfrancopiana “You give it a command that measures something” what did you measure?
English
Max Cagr
11 posts
@_neilstephenson @steipete The ingredients to make Diet Coke existed for ages before @realdonaldtrump actually invented it. What other kind of mixer is chemically possible that nobody’s invented yet?
English
The technology to build OpenClaw existed for ages before @steipete actually built it. What else is sitting there, technically possible, that nobody's built yet?
English
If you haven’t woken up to this, you’re literally going to be left behind.
ₕₐₘₚₜₒₙ@hamptonism
Uranium will be the biggest artificial intelligence trade in history.
English
I am reminded of a posting I saw in a chess forum where some newb with a 700 rating outlined his plan to become a grandmaster in 18 months:
month 1: study openings ( +100 ELO)
month 2: stop making errors (+200 ELO)
month 3: study tactics (+150 ELO)
...
Rogue | Frontier Philosophy@RoguesPhilo
When I say "organize" this is what I mean. 1. Unify your family 2. Generation one: buy amd organize the land 3. Generation two: develop skills and monetize land 4. Generation three: develop a strong local economy 5. Generation four: develop a state-wide supply chain with other self sufficient family farms. It's that simple, yet, no, we cannot encourage families to organize — that's "fed posting" at best. The immediate response to organize is always "You'll be drone strikes out of existence" or "the government will drop napalm on your land and sacrifice your cattle to Satan." It's ridiculous. Americans appear possessed to run a script whenever anyone suggests they can be self sufficient. The script suggests a multitude of arguments about how one cannot move, nor act beyond the technological system's dominion. Like rabid dogs, they spit and they spew and they contort to avoid any personal responsibility to save themselves. It's frankly disgusting. We post into the void for years about the horrible state of our people and yet whenever we are tasked to be our own remedy, we are decried and the strategies denied as heracy. One must simply respond in return "Are the drone strikes against the Amish in the room with us right now, brother?".
English
Kind of crazy that you just need to put 24 words in the right order and a decentralized network allows you access to $111 billion dollars in bitcoin.
We don’t spend enough time focused on this opportunity.
English
@vatsal_manot Interesting. In my opinion, iOS apps are vastly superior to Android and web apps on mobile in terms of polish, cohesion, and fluidity.
English
- Poor scroll performance
- Random glitching of cell-like elements
- Poor text(field|editor) sizing/focus interactions
- Navigation bar titles/items flickering in
- Poor state restoration
… are a few points to start for iOS SwiftUI apps. For macOS the list is much longer.
sam henri gold@samhenrigold
SwiftUI apps all have a certain look and smell to them but I have no idea how to articulate this into my big boy words
English
Active learning is one of the coolest techniques you'll use out there.
If you need to label a ton of data:
1. Take a small portion of the data
2. Label these samples manually
3. Train a model with these labels
4. Use that model to label the remaining unlabeled data
5. Select the most informative samples from the data
6. Label these samples manually
7. Train another model with the labeled data
8. Repeat until your model is good enough
The key to this process is selecting "the most informative samples" using the predictions generated by the model. There are many ways to do this, but generally, we want to pick a combination of samples that increase the diversity and the uncertainty of the data we'll use to train the model.
• Diversity: Identify any unusual, underrepresented, or unknown samples to the model.
• Uncertainty: Identify samples near a decision boundary that are more likely to be misclassified by the model.
You can build a model using Active Learning without labeling the entire dataset, and this model can be as good—if not better—than a model trained with the entire dataset.
Of course, there's no free lunch: getting Active Learning right takes time. Creating a good process to select the "most informative" samples is not always obvious.
English
@personofswag Oh man, when we were in Japan, my girlfriend kept asking me to convert prices for her. One day, I finally explained an easy way for her to do it in her head because she wouldn’t stop asking. Huge mistake, missed it almost immediately. It’s really over for us, isn’t it?
English
