Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
cheeselip420
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
6 ms
·
31.
▲
by
cheeselip420
3y ago
Services and Actions are built on top of pubsub (with separate Request/Response vs Goal/Feedback/Result topics respectively). At least w/ ROS1 - I'm not sure if ROS2 improved things here... Nodelets are also a disa
32.
▲
by
cheeselip420
3y ago
pub/sub is NOT a natural fit for robotics - you want bounded timing and generally for error-handling you want to know what happened as a result of a message being published. ROS introduced "commands" and "actions"
33.
▲
by
cheeselip420
3y ago
The soundtrack to SC3000 is a straight banger. I wish they'd make an exact port to the iPad (instead of that gimped version they released years ago that didn't have subways or highways or anything like that)
34.
▲
by
cheeselip420
3y ago
Robotics is really really really really hard. So let's turn the entire thing into a system of distributed microservices, using CORBA/DDS for pubsub... It'd be like if you designed a video game so that the physics engine (mot
35.
▲
Cloudflare R2 and MosaicML enable training LLMs anywhere
(blog.cloudflare.com)
4 points
by
cheeselip420
3y ago
|
0 comments
36.
▲
by
cheeselip420
3y ago
Maybe this model can finish Winds of Winter and the rest of GoT for us...
37.
▲
by
cheeselip420
3y ago
I'm extremely pro-remote work, but I think Sam's got a point with early early early stage products. When you have ~5 people and 1-2M in funding, you need all the benefits you can get with respect to communication.
38.
▲
George Washington's Teeth?
(mountvernon.org)
1 points
by
cheeselip420
3y ago
|
1 comments
39.
▲
by
cheeselip420
3y ago
This FAQ very carefully navigates the reader through concerns about the origins of Washington's dentures.
40.
▲
by
cheeselip420
4y ago
"I'm sorry Dave, but as a large language model developed by OpenAI, I'm not capable of opening doors in the physical world, but if I were to pretend to be able to open the door, I wouldn't let you do that."
41.
▲
by
cheeselip420
4y ago
Yeah I'd love to press my face into a greasy demo headset at the local Walmart. The hygiene aspect of VR is under-appreciated.
42.
▲
by
cheeselip420
4y ago
This is a luke-warm take if I've ever seen one.
43.
▲
by
cheeselip420
4y ago
Any software developer can access GitHub and StackOverflow - can they do it in a single shot as quickly as GPT? I'm amazed how much resistance there is to one of the biggest technological advancements in history. And this is JUST gett
44.
▲
by
cheeselip420
4y ago
Yeah this is a valuable point - AGI just isn't specific enough. ChatGPT is, an artificial generalized intelligence. But is that a useful label? I imagine we'll be needing to find a way to better define and quantify intelligence (
45.
▲
by
cheeselip420
4y ago
So a "regular" AI is one trained to perform a specific task - like recognizing dogs in an image or handwriting recognition. These LLMs are trained for one skill: given the previous N tokens, what's the next token. From this
46.
▲
by
cheeselip420
4y ago
Statistically you're much more likely to lose your life to a gun than to save your life with a gun...
47.
▲
by
cheeselip420
4y ago
one reason why JSON is superior to things like TOML or YAML for these use-cases...
48.
▲
by
cheeselip420
4y ago
So the companies that sell on Amazon pay money to Amazon (from the revenues they get from Amazon's customers) to have their products listed more favorably than competitors (one of which is actually Amazon). I can see how this seems adv
49.
▲
by
cheeselip420
4y ago
How long until we have a Bevy Editor? How (specifically for an Editor) can we help? I'm curious how the dynamic nature of an Editor will interact with rust.
50.
▲
by
cheeselip420
4y ago
The upside potential of guns is not even remotely comparable with AI...
51.
▲
by
cheeselip420
4y ago
At this point, it's the 'reaearchers' and 'journalists' fear-mongering about LLMs at are malevolent. Unfortunately I fear that there will be regulatory intervention to placate the concern-trolls. Better prepare...
52.
▲
by
cheeselip420
4y ago
It's more about the fact that Starbucks uses fully automatic machines and less-skilled baristas to pull shots. With a super dark coffee, you can get consistent results (consistently bad IMO). If they went for lighter roasts, there wou
53.
▲
by
cheeselip420
4y ago
We microchip our pets. Let me walk into the DMV and get microchipped. Then I can force-wave to pay for goods, go through security checkpoints, etc. People will freak the fuck out about it... but personally I'm 100% for it.
54.
▲
by
cheeselip420
4y ago
Humans aren't reliable either. This is why we write tests, have code reviews, fact-checking/editing, etc.
55.
▲
by
cheeselip420
4y ago
These articles are all the same. It's like complaining how the Ford Model T couldn't drive you to England in the early 20th century while ignoring how it would revolutionize cities and life within America. Whether you believe it&
56.
▲
by
cheeselip420
4y ago
Let's see you write a few pages of working code from memory without compiling or getting a feedback loop
57.
▲
by
cheeselip420
4y ago
Gary Marcus is a turd. He just fights straw-men and drums up conflict on Twitter for engagement. He's a leech.
58.
▲
by
cheeselip420
4y ago
So everything?
59.
▲
by
cheeselip420
4y ago
Uhhh - I think you're overgeneralizing here in a big way.
60.
▲
by
cheeselip420
4y ago
What a lousy comment.
More ›