luke-stanley

8 ms·

1.
by luke-stanley 18d ago
I couldn't quite figure out Lemmalog's latest benchmark stats from the README but it sounds cool! I heard of Scallop, that also uses Datalog for neurosymbolic programming, I wonder where Lemmalog fits with Scallop? I know little a
2.
by luke-stanley 21d ago
Using local models, usually means generating content on a more private, local device with the weights of the model stored on that device - a local LLM model could have helped write these posts. It only gets easier to self-host increasingly
3.
by luke-stanley 3mo ago
But did it Blender open the source?
4.
by luke-stanley 3mo ago
Full visibility of the source code is very different from having full legibility of the system: comprehensibility is the bottleneck. (Seems I ran out of edit time!)
5.
by luke-stanley 3mo ago
The cause of this is that the cost of creating plausible code contributions has gone down, so PR proposals can multiply, but flaws still threaten project security and LLMs can be confidently wrong. So human review is needed right now to mai
6.
by luke-stanley 4mo ago
I got a pocket-sized Bluetooth keyboard (also used for travel around Asia), 195mm x 85mm x 15mm, with a flap - NOT a foldable keyboard, though it does have a cover flap and a kickstand. Sold under a few brands, but the dimensions are quite
7.
by luke-stanley 4mo ago
I'm not sure that copying a key after unlocking the system counts as a backdoor? If the OS promises to lock access to the key and fails to do so then I can see the logic that people might then call that a backdoor. But it's differ
8.
by luke-stanley 5mo ago
Would ADS-B Exchange heatmap files would be accessible at even DEFCON 2? Which raises the question of what it would show when the data sources are not available.
9.
by luke-stanley 5mo ago
Yeah, though it's not great marketing. Especially for hiring interpretability researchers. Their own alignment research has reward model interpretability, personality features and so on (see https://alignment.openai.com )
10.
by luke-stanley 5mo ago
It's a funny detail to skim, but what's more surprising is how mechanistic interpretability and alignment science have much better tools and research than the goblin blog post suggests, including from OpenAI's own alignment t
11.
by luke-stanley 5mo ago
`Is it really going to have ecosystem effects?` Will mandatory ID gatekeeping of developers have ecosystem effects? Surely the only question is how much? You may install apps over ADB every day, but APK installing is much more convenient an
12.
by luke-stanley 5mo ago
But the "opt-out" will not prevent ecosystem effects caused by the default shutdown of convenient app installs due to the policy. Not even for GrapheneOS users. It's a global policy by a body we never voted for. You can'
13.
by luke-stanley 5mo ago
Amazon blocking their agents is addressed by their own Chromium fork running on the users own device.
14.
by luke-stanley 6mo ago
Hey akshayka, I'm not that teacher, but I am a fan, and I notice that marimo sharing would be easier if export and import from normal boring Spyder/Jupyter percent formatted .py files worked by default, instead of indented `@app.c
15.
by luke-stanley 6mo ago
So we need a new standard problem due to the complexity of the last standard? Isn't unicode supposed to be a superset of ASCII, which already has control characters like new space, CR, and new lines? xD
16.
by luke-stanley 6mo ago
Should we really buy the many months of switching difficulty argument? Surely the main API surface is a HTTP API like ChatCompletions? If it's the exact shape of Anthropic's API, the difference is surely minor. There are likely up
17.
by luke-stanley 7mo ago
This is not off-topic at all!
18.
by luke-stanley 9mo ago
Wow. Possible ways to kept Meta ad records honest and transparent: - CCing archive.org - Store on an append-only system with hashing, hello blockchain use-case ha ;D.. IPFS or even GitHub should do, no crypto payments required. - Third-part
19.
by luke-stanley 9mo ago
Currently it says they have released metadata and album art. Is archiving and sharing the textual track metadata alone (no images, no audio) legal in the US, or Europe? By what basis is it legal or illegal?
20.
by luke-stanley 10mo ago
They have this page for reporting: https://github.com/hackclub/hcb/blob/main/SECURITY.md
21.
by luke-stanley 10mo ago
Old man yells at cloud: that's not what "literal killer" means.
22.
by luke-stanley 10mo ago
I had the same reaction as them many months ago, the Google Cloud and Vertex AI stuff namespacing is a too messy. The different paths people might take to learning and trying to use the good new models needs properly mapping out and fixing
23.
by luke-stanley 10mo ago
I'd lower the temperature and try DSPy Refine loop on it (or similar). Using audio encoder-decoder models and segmentation are good things to try too. A length mismatch would be bad. DSPy has optimisers. It could probably optimise well
24.
by luke-stanley 10mo ago
Simon, sorry I didn't get around to answering your question on post-t5 encoder-decoders from the Markdown Lethal Trifecta prompt injection post. ( https://news.ycombinator.com/item?id=45724941 ) Since the plain decoder m
25.
by luke-stanley 10mo ago
I am very picky, hard to place, but from a quick look at the README, I'd say the API interface on display seemed like the right level of abstraction for having to deal with the messy reality. Since you're asking for feedback: - pe
26.
by luke-stanley 11mo ago
Direct quotes: "It pays to be brave." ... "we pledge to distribute to tippers $200 million out of every $1 billion we collect." ... "This is TruthWave. Welcome to the platform and community for those who bring uneth
27.
by luke-stanley 11mo ago
Humans would not do what was reported here!
28.
by luke-stanley 11mo ago
Summarisation models that do not follow instructions already exist! Fixing exfiltration is good, low hanging fruit. But for a summarisation task, whole classes of typical instruction following behaviour are totally off target!
29.
by luke-stanley 11mo ago
The Lethal Trifecta strikes again! Mermaid seems like a bit of a side issue, presumably there are lots of ways data might leak out. It could have just been a normal link. They should probably look further into the underlying issue: unrelat
30.
by luke-stanley 1y ago
Perfectly? Surely that is not reasonable?

More ›