Someone so terrified of having their beliefs challenged that they choose to have none.
They'll be the ones saying that "both sides are equally bad" and think it gives them a moral high ground.
GP specifically mentioned "European radical NATO centrists". They likely think that it's up to Zelensky to stop the war and legitimately can't understand why you don't try to appease people like Putin.
Those sound like reasonable positions, not a moral high ground. Unless you believe Putin will stop the war, it's up to Ukraine to end it.
Or keep it going and send more and more bodies into the meat grinder. Which I wouldn't mind as much if all the bodies walked into it of their own free will, but it looks like everyone voluntarily fighting Russia is already dead. If it wasn't like this Ukraine's "recruiters" wouldn't need to snatch people off the street, which I first saw videos of a long time ago.
Very much this... The utility of Claude has grown exponentially with the amount of work I've put into memory/context management, tooling, skills development and CLAUDE.md (and friends). I freely admit I may just be fortunate, but I've yet to hit a bug in a new release that rose above the level of minor annoyance.
The thing that really validated this point was when I recently tried Codex again. Out of the box it was OK, but not great. I swung over the memory apparatus and had it import directives from my existing Claude install and it was almost immediately quite useful in its own right and, especially, as a counterpoint to Claude when dealing with some fairly complex operational issues.
It depends hugely on what "rent usage of this model through one of the many LLM hosting providers" means. If you're asking them to host the model privately then yes, all of that 1.6T of RAM is likely in use holding weights, activations and KV cache by an inference engine that's only getting/answering requests from you alone. When you aren't actively using the model the hosting process is still active and waiting with all of that memory still wired to it.
As background: For the most part VRAM oversubscription/paging/swapping isn't a thing in the same way that RAM for a VM often is. There are some approaches to it, but (to my knowledge) not at that sort of scale.
There are some systemic reasons for this, but very broadly speaking the GPU vendors are building toward the highest bandwidth and lowest latency possible, and the overhead/complexity of something like protected memory modes serves neither of those priorities.
$5K seems like an absolute steal compared to paying contracted security experts to find such bugs. I'm surprised these programs aren't pushed harder, as the potential ROI seems fantastic.
> $5K seems like an absolute steal compared to paying contracted security experts to find such bugs.
Big companies who run these programs don’t use them as their only security. It’s a bonus program. They have internal employees and often third parties doing testing too.
> I'm surprised these programs aren't pushed harder, as the potential ROI seems fantastic.
I have some exposure to managing one of these programs, which was started by someone who left the company abruptly (right after it went on their resume as an achievement).
It’s a huge amount of work. We got an unbelievable number of bad reports before AI was available. Now it’s just a nightmare. The ratio of bad reports to actual payout-worthy reports was out of control.
Most submitters also tried to inflate the severity of their submission to get to the higher tiers. This created a problem where people were withholding small bugs while they’d try to find a way to chain it or elevate privileges. You could tell because as soon as you detected and closed their exploit, they would hurriedly submit a rushed report to try to claim it and then argue with you for not paying out because it was fixed before they submitted it.
We had payouts that went to larger numbers, but honestly I think it would have been better off if we didn’t. This makes bystanders irate, but honestly the incentives get bad when the payouts are large and everyone is spamming LLM bots at your service.
I do something similar - as I approach the context limits I have a pre-compact flush skill that extracts anything useful from the context, updates the MEMORY.md and my Obsidian vaults (set up as a poor man's graph DB) and so forth. Once everything's been stored I run /compact to keep the general session flow intact. Recently I added a small embedder/vector search setup to the same skill, which seems promising so far.
On another environment I've been doing something roughly similar, but have integrated Hindsight as a kind of all-in-one of the above and am still trying to suss out the best compaction strategy.
The cost issue is obviously of prime importance, but I'd also argue that the transparency of the innovations creates a tremendous cross-pollination, and not only within the Chinese communities but in the US/Europe as well. How many of us are learning the practical aspects of actually running and building AI based primarily on open models? As an example - how far would the work of vLLM or SGLang or even NVIDIA itself (all random examples) be without these models and the challenges they pose?