Twirrim

8 ms·

91.
by Twirrim 7mo ago
Which is why you both listen to what they say, and pay attention to what they do, and what they prioritise. You use the actions to figure out where they were coming from with the message, and then you adapt your message to suit that. &gt
92.
by Twirrim 7mo ago
In the US there is no notice period. In every state (with the exception of Montana), employees are "At Will" only. There is no notice period, or severance pay required by law. Health insurance goes with your job, too, as do any
93.
by Twirrim 7mo ago
Massachusetts used to be one of the most favourable states for non-compete agreements, with strong legal protection and support, favouring companies. Not sure if that has changed since the last time I looked (been a few years).
94.
by Twirrim 7mo ago
I don't think you realise just how cheap and easy it is to run these things. Even at the worst rate of being scraped by AI companies, on the order of dozens of RPS, it didn't even use 1% of a CPU to give them content, nor does it
95.
by Twirrim 7mo ago
I have no idea if it works, but Anthropic in particular spent a lot of time crawling the tar-pit[1] I had running on my domain. They were the reason I set up the tar pit in the first place, as they were at one stage averaging 5 requests pe
96.
by Twirrim 7mo ago
I barely used or remember the ZX-81 my folks had with it's amazing 1KB of memory. It had a 16K expansion module you could plug into the back, which apparently made a big difference, but also didn't have the greatest connection.
97.
by Twirrim 7mo ago
Even worse, explicitly telling it not to do something makes it more likely to do it . It's not intelligent. It's a probability machine write large. If you say "don't git push --force", that command is now part o
98.
by Twirrim 7mo ago
A lot of programs over-allocate on virtual memory, but don't actually use it, and the OS is smart enough to just pretend like it allocated it. I'm sure there's probably some justification for it somewhere, but it's hard
99.
by Twirrim 7mo ago
About 10 years ago now, when what would eventually become Oracle Cloud Infrastructure had just done an internal launch of the first availability domain, as we got ourselves read for public launch late in the year, several senior staff and e
100.
by Twirrim 7mo ago
I guess it's not gambling, or it'd be covered by the UKs existing laws around gambling that set the minimum age to 18. edit: I'm pointing out the UK has apparently decided lootboxes are not gambling, because if they did class
101.
by Twirrim 7mo ago
S3 was well aware of the pain when I was there ~10 years ago, just considered themselves handcuffed by the decisions made before the idea of a cloud was barely a twinkle in a few people's eyes, and even the idea of this kind of scale o
102.
Python: The Optimization Ladder(cemrehancavdar.com)
349 points by Twirrim 7mo ago|146 comments
103.
by Twirrim 8mo ago
unsloth's quantized ones. They mention on the site that this links to that a couple of days ago they released updated freshly quantized versions of Qwen3.5-35B, 27B, 122B and 397B, with various improvements.
104.
by Twirrim 8mo ago
3.5 is doing a good job of reviewing code, even without prompting it to be brief and/or rude.
105.
by Twirrim 8mo ago
I'm getting about 15-20 tok/s with a 128k context window using the Q3_K_S version. For running the server: $ ./llama.cpp/build/bin/llama-server --host 0.0.0.0 \ --port 8001 \ -hf unsloth/
106.
by Twirrim 8mo ago
I've been using opencode pointing to the local model running llama.cpp. The last thing I was having it build is a rust based app that essentially pulls data from a set of APIs every 2 minutes, processes it and stores the data in a loca
107.
by Twirrim 8mo ago
I've been finding it very practical to run the 35B-A3B model on an 8GB RTX 3050, it's pretty responsive and doing a good job of the coding tasks I've thrown at it. I need to grab the freshly updated models, the older one seem
108.
by Twirrim 8mo ago
I had a similar speed jump, went from a Celeron 233 to the Athlon 1Ghz. Such a huge uplift.
109.
by Twirrim 8mo ago
The disdain on my part was very much towards the egotistical tech bros who were convinced they could do better in fields they had no background in. It used to get really tiring seeing the rhetoric about every single field of expertise. So
110.
by Twirrim 8mo ago
The old mental model doesn't fit how any OS manages RAM. Every OS plays all sorts of fun guessing games about caching, predicting what resources your program will actually need etc. The OS does a lot of work to ensure that everythi
111.
by Twirrim 8mo ago
When I was at AWS over a decade ago, there was endless complaints about the elevator algorithms by engineers, with the usual egotistical tech-bro insistence that they could do better. Things used to really suck at lunchtime when folks woul
112.
by Twirrim 8mo ago
I've been testing the same with some rust, and it's has spent a fair bit of time going through an infinite seeming loop before finally unjamming itself. It seems a little more likely to jam up than some other models I've exp
113.
by Twirrim 8mo ago
I haven't seen a page on HF that'll show me "what models will fit", it's always model by model. The shared tool gives a list of a whole bunch of models, their respective scores, and an estimated tok/s, so you
114.
by Twirrim 8mo ago
I took a quick look, the dependency on php 8.5 is mildly irritating, even Ubuntu 26.04 isn't lined up to ship with that version, it's on 8.4.11. You mention in the README that the goal is to run things in a standard environment, b
115.
by Twirrim 8mo ago
Hell, the number of times I've used a lot of the data structures that come up in leetcode exercises without at least looking at some reference material is pretty small. I usually assume I'm going to misremember it, and go double
116.
by Twirrim 8mo ago
> reaching to the point of saving php code into the mysql database and executing from there. Wordpress loves to shove php objects into the database (been a good long while since I used it, don't remember the mechanism, it'd be
117.
by Twirrim 8mo ago
> The network latency bit deserves more attention. I’ve been trying to find out where AI companies are physically serving LLMs from but it’s difficult to find information about this. If I’m sitting in London and use Claude, where are the
118.
by Twirrim 8mo ago
I've been fighting with an AI code review tool about similar issues. That and it can't understand that a tool that runs as the user on their laptop really doesn't need to sanitise the inputs when it's generating a comman
119.
by Twirrim 8mo ago
I've seen stuff posted about chained assignment footguns in python regularly over the years, and it always surprises me. I don't think I've ever written them, or reviewed code that does. I don't think it'd occur t
120.
by Twirrim 8mo ago
Just in case anyone hasn't seen this yet: https://github.com/ggml-org/llama.cpp/discussions/15396 a guide for running gpt-oss on llama-server, with settings for various amounts of GPU memory, from 8GB on

More ›