hedgehog

6 ms·

91.
by hedgehog 3mo ago
I do essentially all of my work with auto compact set to 250k and it's fine. It may be due to the way the project tooling is set up and the use of sub agents?
92.
by hedgehog 3mo ago
What models have you found to work for which tasks? I find the local models very useful but not in a way that's replaced cloud models (yet, I remain hopeful).
93.
by hedgehog 3mo ago
I don't know the details but I would imagine that the comparison would shift quite a lot over those years due to the rollout of SIMD features on both sides along with the other microarchitectural improvements. I'm not sure I would
94.
by hedgehog 3mo ago
Pentium 133MHz was coincidentally the fastest machine I had at the time but also the minimum spec labeled on the box. (my laptop had a Chips & Technologies 65550, so no 3D help there)
95.
by hedgehog 3mo ago
Yes, but Half-Life was designed to be somewhat playable with software rendering even on a 133MHz Pentium and that would have been the better rendering path on the earliest iMacs. In my opinion in that generation Half-Life's software re
96.
by hedgehog 3mo ago
At the time there was also a major push to bring desktop computing and the web to the general population, remember this was a time when cutting costs enough to market a $1000 PC ($2000 adjusted for inflation) was considered a major achievem
97.
by hedgehog 3mo ago
For reference this is roughly period correct for the early iMac models (G3 and DV) that came out in 1998 and 1999, though MacOS 9 didn't ship until later. Impressive porting work.
98.
by hedgehog 3mo ago
This is SSH into a remote Linux VM that has the GPU attached.
99.
by hedgehog 3mo ago
They do reset weekly quotas periodically, I'm not sure how they determine when.
100.
by hedgehog 3mo ago
The whole point the guy is making in the transcript is that they're taking a different strategy from the US labs, one where they focus on smaller models and cost control, and maintain as top priority the work stream that they think wil
101.
by hedgehog 3mo ago
It's probably getting less discussion because there's less user energy, the existing VM / database product has clearly not been a top focus for a while and in any case that business is becoming even more of a tough commodity
102.
by hedgehog 3mo ago
Maybe enabling MTP causes some weights to be displaced to host memory? MTP itself doesn't do anything during prefill so that should be exactly unchanged, decode will vary depending on settings but with 2-4 proposals depending on worklo
103.
by hedgehog 3mo ago
MTP has no effect on prompt processing.
104.
by hedgehog 3mo ago
Yes that's what I've read. As far as I know the approach should transfer well to hybrid model architectures like modern Qwen and sizes like 27B by using multiple chips. LoRA-steered Qwen 27B at 10K+ tokens per second would be tran
105.
by hedgehog 3mo ago
No the weights are in the metal layers, they cannot be updated.
106.
by hedgehog 3mo ago
The library does respect the client-side reduce motion setting (by disabling all animated effects).
107.
by hedgehog 3mo ago
For our purposes it has the mining/crafting/construction bits the kids care about so in that dimension there's no difference. From my perspective it's been simple to operate, and having all the source makes a good base f
108.
by hedgehog 3mo ago
Luanti with Mineclonia plug-in works well if you are willing to move entirely to Mac/Windows/Linux clients. The whole stack is open source so it's easier to make changes etc if you want. No hacky binary patching of Java bytec
109.
by hedgehog 3mo ago
The VAIO P and some Panasonic rugged handhelds have those chips, both interesting machines, I think it would be possible to write a modern Mesa driver for those but it would be a lot of work.
110.
by hedgehog 3mo ago
I auto compact around 200k tokens both due to this and because the cached read cost really escalates when sessions have more tokens than that (too short and you pay a lot in per-compact re-reading of state)
111.
by hedgehog 3mo ago
I don't know about the overall breakdown but in my case longer runs are prototyping, bug hunting, reverse engineering, etc. For example Gnome Remote Desktop didn't work in my configuration due to a combination of hardware and code
112.
by hedgehog 3mo ago
I don't think the current subscription price is intended to be a money maker. It's the loss leader to get people invested in the companies' tooling, and make those people more willing to start paying higher enterprise rates a
113.
by hedgehog 3mo ago
It's what some more experienced users want and the companies are following the well-trod path of optimizing heavily for power users at the expense of complexity, only now it's gotten easier to add absurd amount of code to a projec
114.
by hedgehog 3mo ago
Once native apps became important Microsoft didn't have a chance, the basis for Apple's success was in place back in 2003-2005. Apple had a big head start on what is now the standard phone template (always-on cell data, capacitive
115.
by hedgehog 3mo ago
For the sake of accuracy Claude Code has a "fork" feature which does what you'd expect and based on traces I skimmed this morning it seems under some circumstances you get fork behavior for subagents as well.
116.
by hedgehog 3mo ago
If you are not issued body armor and K&R insurance FDE seems like the wrong term. (the use of "engineer" for non-PEs is... a fair debate)
117.
by hedgehog 3mo ago
I don't think that's quite accurate, the screens of the time were more like 480px wide and I don't think most people had a strong aesthetic opinion. Just look at Android. The major problem is that whatever app you cared about
118.
by hedgehog 3mo ago
It wouldn't necessarily have been better, a major reason the Windows phone stuff failed is it didn't have market share to justify app development. Android barely made it work as a well-funded #2. Palm WebOS, MeeGo, there were vari
119.
by hedgehog 3mo ago
That is true of Anthropic's implementation but not inherent in sub-agents in general.
120.
by hedgehog 3mo ago
I don't have the experience, how high are Haskell compile times such that a switch like this is worth it? I have some experience with generated Python, Rust, and TypeScript, and I have not found that compile time is enough of a concern

More ›