sorted by: new top controversial old
[Andrew Parsonson] DLR to Purchase 80 Red Kite Sounding Rocket Motors - European Spaceflight in c/spaceflight@sh.itjust.works
[Video] U.S. military air defense systems seen deployed around President Trumps Mar-a-Lago estate in Palm Beach after rumors of strikes on Iran in c/politics@lemmy.dbzer0.com
Day 831 of posting a Daily Screenshot from the games I've been playing in c/games@lemmy.world
Citroen revives the 2CV for an electric age in c/electricvehicles@slrpnk.net
What do you use for alerting? in c/selfhosted@lemmy.world
mold version 3.0.0, rewritten in Rust in c/rust@programming.dev
Ollama's new "Nimble" model is a 9B decision engine, not a chatbot — and it's fast in c/localllama@sh.itjust.works
[–] 1 point 1 month ago*

While you can, there has been research that showed that the LLMs answers can be influenced by the order in which the possible options are presented.

I haven't looked into how these "System One" "Noul" models (absolutely stupid nomenclature imo) work but the hope would be that theyre less susceptible to such problems.

  • permalink
  • fedilink
  • source
  • parent
  • context
    • xkcd #3305: Ground Effect in c/xkcd@lemmy.world
    NASA mysteriously trying to pull world’s fastest plane, SR-71 Blackbird, out of retirement — and won’t say why in c/technology@lemmy.world
    [–] 4 points 1 month ago (1 child)

    The thing is sounding rockets do exactly that and are still in service (vs somehow unfucking an ancient plane that hasn't been serviced in decades with tooling that no longer exists) and there are multiple companies out there operating supersonic drones for high speed tests of whatever you wanna do.

    I dont see the value in bringing the blackbird back into service. I think its more likely that they want to freshen up the paint and park it at some press conference / event.

  • permalink
  • fedilink
  • source
  • parent
  • context
    • Meet the TD110 Maelstrom in c/helldivers2@lemmy.ca
    September 15, 1966, after a three-day, 44-orbit journey in space, Gemini 11 astronauts Pete Conrad and Dick Gordon splash down in the Atlantic in c/historyphotos@piefed.social
    [–] 5 points 1 month ago* (2 children)

    A little over 20 mins:

    Splashdown in the western Atlantic at 24.25° N, 70.00° W, 4.9 km from the target point, occurred at 8:59:35 a.m. EST. The crew was picked up by helicopter and brought to the U.S.S. Guam at 9:23 a.m. and the spacecraft was recovered at 9:58 a.m.

    Source

  • permalink
  • fedilink
  • source
  • parent
  • context
    • Steam Programming Fest is here! in c/steam@lemmy.ml
    25
    Authentik Helm woes (discuss.tchncs.de)
     

    I'm currently in the long process of rebuilding my declarative homelab using k3s, ArgoCD and NixOS.

    I had previously used Keycloak but that always seemed massively overqualified and way too complex for my purposes. With this rebuild I saw my chance to try out Authentik which appears to be in good standing with the homelab community.
    They have tons of documentation for pretty much anything which was encouraging to me. Well except for the documentation for their Helm Charts maybe...

    Started off with version 2025.12.x, am now onto 2026.02.x and have spent most weekends in between that on getting Authentik to even just deploy to the cluster.
    It's partially my fault for attempting to use Secrets initially but even now with hardcoded keys in my git repo the default example chart doesn't work:

    values.yaml

    authentik:
  existingSecret:
    secretName: authentik-secret

  postgresql: # None of this gets applied at all so I do it manually below...
    password: "somepasswd"

server:
  replicas: 1

  env: # Manually apply all the configuration values. Why am I using Helm charts again?
    - name: AUTHENTIK_POSTGRESQL__HOST
      value: authentik-postgresql
    - name: AUTHENTIK_POSTGRESQL__USER
      value: authentik
    - name: AUTHENTIK_POSTGRESQL__PASSWORD
      value: "somepasswd"
    - name: AUTHENTIK_POSTGRESQL__NAME
      value: authentik

  route:
    main:
      # ...

postgresql:
  enabled: true

  auth: # And set everything here once again
    username: authentik
    password: "somepasswd"
    postgresPassword: "somepasswd"
    usePasswordFiles: false
    database: authentik

  primary:
    persistence:
      size: 4Gi

    I started off with the official example and after all these undocumented changes it still only deploys-ish:

    With the defaults authentik-server would always try to reach the DB under localhost which doesn't work in the context of this chart/k8s.
    So after a while I figured out that the authentik: configuration block doesn't actually do anything and I set all the values the chart should set by hand.

    Now the DB connects but the liveliness probe on the authentik-server pod fails. It logs the incoming probe requests but apparently doesn't answer them (correctly) leading to k8s killing the pod.

    Sorry for the ramble but I've hit my motivational breaking point with Authentik.
    Since the community seems to like it a bit I am left wondering what I'm doing wrong to have this many issues with it.

    Did you people have this much trouble with Authentik and what have you switched to instead?

    17
    The release of asciinema CLI 3.0! (blog.asciinema.org)
     

    cross-posted from: https://lemmy.ml/post/36210816

    asciinema (aka asciinema CLI or asciinema recorder) is a command-line tool for recording and live streaming terminal sessions.

    This is a complete rewrite of asciinema in Rust, upgrading the recording file format, introducing terminal live streaming, and bringing numerous improvements across the board.

    view more: next ›