HN user
[ my public key: https://keybase.io/mwarkentin; my proof: https://keybase.io/mwarkentin/sigs/W8W1ADGNvjCb10QTKZCkSneAxKTKSObDYwEoFzpC0SM ]
Just gives you an error if you have a link to a “news” site in your post.
News content can't be shared in Canada In response to Canadian government legislation, news content can't be shared. Learn more (links to https://m.facebook.com/help/2579891418969617/)
https://cbc.radio-canada.ca/en/media-centre/blocking-of-news...
Not open source? :(
Is it? If the spec is as detailed as the code would be? If you make a change to one part of the spec do you now have inconsistencies that the LLM is going to have to resolve in some way? Are we going to have a compiler, or type checker type tools for the spec to catch these errors sooner?
Or something like https://antithesis.com/
Ah yes, all of the precedents and lawful authority that this president cares so much about adhering to.
Crypto shredding?
Some have done it (repo includes links to main articles covering the process): https://github.com/getsentry/sentry-watchdog
Note: I work for sentry.
If you’re on iOS, and don’t use swipe keyboard, disable that functionality. I found it made a big improvement for me.
There is a whole KIP that is “preview” on Kafka 4.1 to handle this use case natively: https://cwiki.apache.org/confluence/plugins/servlet/mobile?c...
Note: I haven’t had a chance to test it out in anger personally yet.
“I can stop any time I want!”
Uv solves this (with some new standards). ./script.py will now install the python version, create a venv, and install dependencies (very quickly) if they don’t exist already.
#!/usr/bin/env -S uv run --script # /// script # requires-python = ">=3.12" # dependencies = [ # "ffmpeg-normalize", # ] # ///
Specifically, the paper estimates that Llama 3.1 70B has memorized 42 percent of the first Harry Potter book well enough to reproduce 50-token excerpts at least half the time. (I’ll unpack how this was measured in the next section.)
Interestingly, Llama 1 65B, a similar-sized model released in February 2023, had memorized only 4.4 percent of Harry Potter and the Sorcerer's Stone. This suggests that despite the potential legal liability, Meta did not do much to prevent memorization as it trained Llama 3. At least for this book, the problem got much worse between Llama 1 and Llama 3.
Harry Potter and the Sorcerer's Stone was one of dozens of books tested by the researchers. They found that Llama 3.1 70B was far more likely to reproduce popular books—such as The Hobbit and George Orwell’s 1984—than obscure ones. And for most books, Llama 3.1 70B memorized more than any of the other models.
AI posts (including this one) are all over his employers blog lately, so there’s some stake (fly MCP, https://fly.io/blog/fuckin-robots/, etc).
Yes, some think it’s happening now: https://www.aisnakeoil.com/p/is-ai-progress-slowing-down
FWIW there are protections for this now: https://aws.amazon.com/blogs/compute/aws-lambda-introduces-r...
Probably talking about something like https://antithesis.com
It doesn't look that hard: https://youtu.be/vuvckBQ1bME?si=eCGILMLlZHZnztCR
Yes.
Although it’s also hollowed out the group of people using StackOverflow (and perhaps stack overflow has restricted open access to its data for further scraping), so future iterations of LLMs will have less up to date training data to use.
Back when we first tried out DMS there was a fun bug where booleans were interpreted as strings, so all false records were interpreted as “false” and translated to true on the target DB. It was fixed shortly after but was a good reminder to validate your data during a migration.
I leave my laptop plugged in 95% of the time so in practice this caps it at 80% (helpful).
I suspect quite a few people do that so it probably is net-good overall, but agree it doesn’t work well if you’re moving around a lot.
Yes, I think that's about right (SRE at Sentry).
Pretty sure we had people spin up a Slack instance at our company while we were still on Hipchat officially. Slack was a big improvement over what we had before that IMO.
The Functional Source License (FSL) is a source-available non-compete license that converts to Apache 2.0 or MIT after one year.
Uh, reference for s3 prices going up? I’ve only ever seen them heading down over time.
Yep, we just had a cabinet with a bunch of hardware keys for anyone to take.
Remote employees take a bit more support though.
We're using their "stable" release channel which means most of our clusters are currently on a version of 1.25. We only have a few deployments using PVCs so impact is pretty minimal anyways for us as far as I can tell.
There's rust stuff too: https://github.com/getsentry/relay
Looks like 2.7% of their deposits were > 250k. So yes, it seems relatively unique (there are only 5 banks on this list with <15%).
https://twitter.com/GRDecter/status/1634208652595699713?s=20