hedgehog
Both devices have integrated memory, so that 16 GB will look more like a 11/5, 12/4, or maybe even 14/2 split. The Steam Deck is also $400 for an LCD model or $550 for the OLED, not $800. It’s reasonable to expect more performance when you pay more.
Because the Steam Deck has a lower native resolution, that means that less of the RAM will be used for the integrated GPU. Downscaling from 1080p to 720p doesn’t look good, either - and you could downscale to 540p if supported, but if you need to do that (vs choosing to for an emulated game) it probably won’t be pretty, either.
This device is also running Windows, rather than a streamlined Linux-based launcher, meaning that more of that RAM will be taken up by OS processes by default.
The article talks about how the 8840U benefits from more, fast RAM. You won’t get near the 8840U’s full potential gaming with 16 GB. 24 GB, on the other hand, would have been enough that games expecting 16 GB of system RAM would have been able to get it, even while devoting 6-7 GB to the GPU and 1-2 GB to the OS.
Unless something has changed, it did. The page linked reads:
And, obviously, this POC is open source, the code is publish here on our forge.
The link takes you to their repos. The server repo has instructions on self-hosting directly on your server or with Docker. The app repo has code for both the iOS and Android apps. That’s good, because the iOS app at least doesn’t have a built-in way to select a different backend server.
Whisper is by OpenAI and as far as I know they have not shared the training code, much less the data sets, so the best you can do is fine-tune the models they’ve provided.
If use of Whisper is a problem, but the project is otherwise interesting to you, you could ask them to consider using a different STT solution (or allowing the user to choose between different options). I’m not aware of any fully open STT applications that are considered to be as capable as Whisper, but if you do, that would be great info to share with them.
Realistically, no LLM that’s large enough to be competitive will be able to remain open-source, even if it was initially (and most that claim to be weren’t actually, as you point out), because so much training data is needed.
Often the training data can’t be re-distributed in the first place, but even if it can be, its availability makes it much more likely that someone will request the takedown of some data in the set (even if the data was licensed, someone who holds copyright might claim that the person who submitted it to the set wasn’t permitted to do so). At that point, unless the takedown request is refused or the model itself is re-trained (which would be quite expensive) the data is no longer sufficient to generate the model.
Depends on your perspective. Would it be fine for Meta Threads to replace it? Threads supports ActivityPub, so in some ways it likely interacts better with the fediverse.
If we agree that Threads isn’t a suitable replacement, then clearly there’s some criteria a replacement should meet. A lot of the things that make Threads unpalatable are also true of Bluesky, particularly if your concern relates to the platform being under the control of a corporation.
On the other hand, from the perspective of “Twitter 2.0 is now a toxic, alt-right cesspool where productive conversations can’t be had,” then both Threads and Bluesky are huge improvements.
As it is, you only see new comments if you scroll past the post again (and your client has refreshed it) or if you open it directly. If your client hasn’t updated the comment count or if you refresh your feed and the post falls off, you’ll never see it anyway.
A “Watch” feature would solve this better. If you watch a post, you get aggregated notifications for edits and comments on the post. If you watch a comment, you get aggregated notifications for replies to it or any of its children.
By aggregated notifications, I mean that you’d get one notification that said “The post you watched has been edited; 5 new comments” rather than a notification for each new comment.
Then, in addition to exposing a “Watch” action on posts and comments, clients could also enable users to automatically hide posts that are watched, either by marking them as hidden or by hiding watched posts without updates.
If the latter approach were taken, notifications might not even be necessary - the post could just get added back into the user’s feed when changes were made. It would result in a similar experience to forums, where new activity in a topic would bump it to the front, but it would only impact the people who were watching it.
You can kinda get that behavior by sorting your feed by Active, but this could be used with other sorting methods.
The way to do this is to use a mailing list that only allows a limited number of people to send emails to it. You could do this automatically when someone clicked a “Prohibit Reply All” button, but such a feature is unnecessary if you use mailing lists configured that way by default.