• 16 Posts
  • 1.58K Comments
Joined 1 year ago
cake
Cake day: June 23rd, 2024

help-circle



  • I like that none of my local devices are externally addressable unless an outgoing connection has been established.

    This can also be achieved using (other) firewall rules.

    but then it’s essentially just maintaining a NAT table without the translation piece.

    So… a firewall?

    NAT isn’t a security feature and shouldn’t be relied on for managing access to hosts.

    It also breaks the assumption of IP that connections between hosts are end-to-end, which requires sophisticated solutions so that everything works (more or less).

    I too employ NAT to make services accessible over IPv4. But only because it doesn’t work otherwise. Not because it “makes sense”. I don’t use it at all for IPv6.















  • It’s a weird case. As the paper says, this is inherent to LLMs. They have no concept of true and false, and rather produce probabilistic word streams. So is producing an untrue statement an error? Not really. Given these inputs (training data, model parameters and quiet), it’s correct. But it’s also definitely not a “hallucination”, that’s a disingenuous bogus term.

    The problem however is that we pretend these probabilistic language approaches are somehow a general fit for the programs they’re put in place to solve.