HN user

chillfox

2,963 karma

Sysadmin turned software developer

Posts4
Comments1,079
View on HN

I have been testing the various models, and I would not call Fable SOTA.

I can't actually get Fable to do anything. I only work on back-end code, and the moment Fable notices the jwt scope checks on the endpoints it's game over, it refuses to do anything because security is involved.

So for me, Fable is completely useless, the bar is very low, any llm that will actually attempt the task beats it every time.

Thanks for building this, I am just about to give it ago with my self-hosted Garage cluster.

Does running `stat` against a file require pulling the whole file from s3, or can that be handled by the metadata?

Do you know what backup performance is like for something like borg/borgmatic or restic, especially on follow up runs where most files are just checked.

Is there any particular Redis/Valkey config you recommend when using it for `conditional_put`, or just default config?

Is there any chance for NFSv4 support?

For sure, I am assuming they spy on a lot more people than they drop bombs on.

I remember a long time ago it came out that the US had been doing mass spying on the Danish people, my dad was very upset about it and disliked the US for the rest of his life. Of course the only thing he did about it was not watch American movies anymore or visit the US.

Anyway, I assume it will be a case of a million little paper cuts, each thing putting off a group of people until someday it adds up to real meaningful economic impact.

Podman v6.0.0 20 days ago

They really upset me with how they introduced it initially, and I haven't looked at it since.

Redhat swapped in podman and removed docker, and all of the config/images/scripts I had spend 2 years making stopped working on new/replacement hosts... So I had to spend a lot of time in meetings/filling out paperwork to get docker back. I haven't forgiven Redhat for dropping that shitty experience on my head.

Immich 3.0 20 days ago

I only wish it would support nested albums (or albums in folders) so it could be an easy replacement for lightroom cloud as well.

I have all of my photos organized like this: `events -> year/month - holiday -> (album_1, ...)`. and: `home town -> year -> (album_1, ...)`. Photos will be in multiple albums, and there will be edits as well. And I need to track the picked/rejected state as well (and filter on it).

Only reason I haven't moved over to Immich yet is because I am struggling to map my photo organization onto it's way of doing things in a way that's nice. So far my attempts have been unwieldy.

I worked at a place that argued that nobody would game the metrics because it would be wrong and they never stated what the metrics were… while I was gaming the metrics and they were praising me for being one of the best on the team.

It was an unreal experience.

lol, no!

Having written bots several times, any kind of friction or payment on the json api would make me just use the free html "API" it's just easier.

I have many times used a webpage as api instead of the actual api because using the actual api required doing paperwork, like writing business cases, filling out approval forms, creating accounts, paying, etc...

So... how's that any different from using American stuff for those of us in the rest of the world?

Over the last decade, the US has been way more unreliable than China. There's been a near constant negative impact from the US doing something.

At least with China, we are very good at winning trade wars with them here in Australia.

Why not?

Mythos level really doesn't seem that scary. And it would be a great way to take away the American labs international market.

I think it would make strategic sense for them to release more capable models than what American labs are allowed to make available to the world. It would help them grow their global soft-power and be a destabilizing effect on the American economy.

Fugu Ultra [0] is not actually a model, it's a system (harness in the cloud?) that routes to several models, looks like it's a bit like OpenRouters Fusion [1].

  "Rather than a single monolithic model, Fugu is a learned multi-agent orchestration system: a language model trained to route tasks across a swappable pool of underlying models and to recursively call instances of itself." - https://openrouter.ai/sakana/fugu-ultra

[0] https://sakana.ai/fugu/ [1] https://openrouter.ai/openrouter/fusion

Pretty sure the secret sauce is in the summarised thinking. Maybe better though process… But I have a feeling it’s server side tools and a scratch space to prepare the reply.

Sometimes the summarised thoughts include stuff that makes no sense unless it’s got a workspace on the server. Stuff like “I am now writing x to file y”.

I can't build workflows on something that can randomly be unavailable for over a week.

At this point the future availability of Anthropic models outside the US is very unclear.

Can't do that.

If they get rid of R&D, then someone else will make a better model and we will all switch to using that model.

If sales & marketing covers subsidies and bribes then they cant get rid of that either. Get rid of the bribes and they will be shut down. Get rid of the subsidies and we will all switch to someone cheaper.