Hermes supports this too, just do `/handoff telegram` and your current session is moved to telegram, it pings you on your phone, etc. It's pretty nice, though I do still prefer the Codex dedicated app for remote control
944 karma · joined November 24, 2014
> And so it is that you by reason of your tender regard for the writing that is your offspring have declared the very opposite of its true effect. If men learn this, it will implant forgetfulness in their souls. They will cease to exercise memory because they rely on that which is written, calling things to remembrance no longer from within themselves, but by means of external marks.
That's solvable though, whether through changing training data or RL.
CBC is definitely left wing[1], with a bias towards the liberal party (centre-left). Globe and Mail is centrist[0].
Should bosses also get one free insult for their employees too? Obviously not.
Whether typechecking or unit tests is "better" is really a question of taste.
Yes, and the flow of future models may dry up, but the current local models we'll have forever.
> If an API delivers very solid results one day and crap the next and I spent a lot of money, how does that work? There are many people on reddit/youtube speculating why claude sometimes responds like a brilliant coder and sometimes as if it had a full frontal lobotomy. I see this in Cursor too.
This seems like an incredible over-reach. There's no predatory behaviour here. You're free to cancel at any time.
It's an incredibly fast moving field, a frontier field in software. To say that, in order to charge for something, you are legally bound to never make mistakes and have regressions, is an incredibly hostile environment to work in. You'll stifle growth if people think experiments might come with lawsuits unless they're positive it leads to improvement.
If they decided they were going to lock everything to gpt-2 and refuse to pay back any people who bought yearly subscriptions, sure I would be agreeable to considering this a bait-and-switch hoodwink. But that is clearly not happening here.
But there are a ton of models I can't run at all locally due to VRAM limitations. I'd take being able to run those models slower. I know there are some ways to get these running on CPU orders of magnitude slower, but ideally there's some sort of middle ground.
There are minor downsides in theory. Proof-of-stake is a bit less "democratic" and gives more voting share to those who have more money, vs. just are mining. In practice, proof-of-stake coins seem to be doing fine. The main one here is Ethereum.
The main reason Bitcoin hasn't is due to inertia and the interest for miners (who do proof-of-work) to use their democratic power to keep bitcoin invested in proof-of-work.
The United States declared war on Al-Qaeda, and killed it's leader in response to an attack on 9/11 that leader claimed credit for. Domestic courts don't factor in here. Any country is lawfully allowed to kill enemy combatants or military leadership in a war.
It is because not everyone sees Israel as committing war crimes. I'm not looking in getting into a debate about it, I don't think this is the appropriate thread and do not want to run afoul of HN guidelines. But to treat your question as if it is serious and sincere (i.e. not a rhetorical question) the answer is because many do not see Israel as committing war crimes but basically everyone agreed China was censoring. I'd also add that censorship can be a particularly sensitive topic the tech community is largely more polarized against than the wider country, whereas weapons of war are viewed by some as valid defensive work. Basically, nerds care about censorship disproportionately. And of course, opinions on both China and Israel would be baked into this too.
Again, not looking to get into the actual substance of these points here, merely answering your question for why large amounts of people see these as different.
> In 1963 [edit: 6 years before], the United States Air Force started a series of study projects on a very large strategic transport aircraft. Although the C-141 Starlifter was being introduced, officials believed that a much larger and more capable aircraft was needed, especially to carry cargo that would not fit in any existing aircraft
> In 1965, Lockheed's aircraft design and General Electric's engine design were selected for the new C-5 Galaxy transport, which was the largest military aircraft in the world at the time.[1] Boeing carried the nose door and raised cockpit concepts over to the design of the 747.
And from some other articles:
> At Boeing, Sutter worked on many commercial airplane projects, including the 367-80 "Dash 80", 707, 727 and 737. He eventually became a manager for the new jumbo-sized wide body airplane, the four-engine Boeing 747. [0]
My only point is they didn't go from zero to 747 in 3 years. They went from having an active culture of building jets, and doing R&D in this area, and having people that had successfully worked on other new jet design projects. They went from that to having a 747 in 3 years. Still incredibly impressive, but if we admit that culture is lacking now, it's entirely conceivable it'd take longer than 3 years to build a comparable jet today. (Look at how much slower north america builds rail now, or anything, compared to the 60s).
My non-credible idea would be to just use an Apple-style magsafe charger on the back of the airplane that disconnects midair at 30,000ft and falls on the helpless people below.
If you’re making the point that America and NATO could join the fight sure, but that’s a hypothetical and the reality now is that the skies in Ukraine are contested.
Stringers are absolutely useful in LSCO. Soviet doctrine made heavy use of MANPADS (man portable air defense systems, of which Stinger counts) when they were planning for LSCO. Basically every army besides the US army (and its allies) makes heavy use of them, and the US army is different because they plan for the US Air Force to be able to clear the skies for them. Soviet doctrine embedded air defence at all levels of their army, from MANPADS to Pantsirs, Buks, and S-300s etc. This goes from man carrier to short/middle/long range.
Why would it be popular to nerf tools? I thought people all preferred the more powerful LLM models.
It's unclear what the rules are going to be for AI generated art. There has been no law, or no court case, deciding it. As such, Valve is being a bit more defensive here to limit exposure.
I agree it's not a great trend, but it seems inevitable when this can lead to massive risk (lawsuits including Valve), and the upside is negligible. Valve is not going to lose market share over this, at least not for the foreseeable future.
With this step, Valve massively reduces their legal exposure. Even if some AI are games slip through and the worst case scenario manifests it is still likely Valve will be fine, as they’ve sufficiently isolated themselves from the risk.
As for whether NMS counts, the answer is no. This isn’t about AI per se, it’s about using models trained on other copyrighted works. NMS doesn’t do that in any of its procedural generation.
As for how much AI generated art counts, this is a ship of Theseus problem. Not saying it’s easy, but hardly the first time regulation has come up against it.
> Any organization that designs a system (defined broadly) will produce a design whose structure is a copy of the organization's communication structure. [0]
But I think OP is implying the opposite, sort of a reverse-Conway. That the narrative of their product affected the structure of the company.
I used to use sonarr and radarr in the past but used those directly, not via Plex, and to be honest I sometimes ran into issues with them but generally they worked fine.
I think it's fair to call Heroku (and 12factor apps) a pre-cursor to edge computing, but certainly not an early player.
[0] https://web.archive.org/web/20201126133614/https://devcenter...
Agreed on increased complexity, but isn't the point largely the dynamic queries? You setup all this extra architecture, but in response you don't have to write as many "controllers" with custom API response. Just let the frontend request what it needs via GraphQL. So it's more upfront work, less down the line. Or is writing all the API level sanitizers and query resolvers more work?