RandomBK

Potentially dumb question: wouldn't the existence of a global magnetic field already imply the presence of a global electric field? Or is this discovery a separate unrelated electric field?

I've no plans for it myself, but that sounds like an interesting project! It should be fairly easy to hack together. The actual logic itself is just a few dozen lines of glue code around a single HTTP POST request.

Already exists :)

Set the `local_uri` setting in the configuration (`llm2sh --setup`), and either pass `-m local` on the CLI or set `"default_model": "local"` in the config.

PRs are always welcome.

I'm glad you like it! The whole point is to have a simple no-frills clean CLI experience.

> Is there a way to point at an arbitrary API endpoint?

Yes. Set the `local_uri` setting in the configuration (`llm2sh --setup`), and either pass `-m local` on the CLI or set `"default_model": "local"` in the config.

Were you there for fun, or research? What made you decide to visit in person?

From what I can tell, Mauritania is not particularly tourist-friendly. It has an active "Reconsider Travel" advisory with the US State Dept, and there was a coup there as recently as 2008.

How does one even go about planning the logistics of such a trip?

I'm starting to see some packages break due to not pinning 1.x in their dependencies. `pip install numpy==1.*` is a quick and hacky way to work around those issues until the ecosystem catches up.

I think GP meant <Author>, <Reviewer 1>, <Reviewer 2>

How is this not a blatent antitrust violation?

It's a petition early on in the case, so it just needs to be based upon "information and belief". The fact-finding step of the multi-year process has yet to begin, but lawyers need to present the best plausible case possible at this stage.

I'm pretty sure the Death Star undid several eons of environmental conservation efforts on Alderaan.

Yikes... In the same doc:

>"For example, the separation between kernel mode and user mode is a classic and straightforward security boundary."

>"Non-boundaries: [...] Administrator to Kernel [...] Administrative processes and users are considered part of the Trusted Computing Base (TCB) for Windows and are therefore not strong isolated from the kernel boundary."

The reasoning makes sense, but that's quite a mixed message for an unintuitive nuance.

Thanks, I hate it.

* I love the idea, and I'm interested to see more progress in this direction.

* I'm not a huge fan of the syntax - for me, the commands are the star of the show, while the async/await is a glue. This syntax emphasizes the async/await while the important stuff get shuffled into quotes.

* Don't make me install node - and especially not system/user-wide packages - to run a script.

Having said all that, this is a neat idea and a cool hack on top of javascript language features.

My job does not involve direct responsibility for the immediate life-or-death of hundreds of lives.

Alternatively, the technique used for finding plane crashes has a 99% recall and we're just happening to talk about the 1% that slipped through. Survivorship bias is not out of the question here.

They're even more terrified of torch or another framework take up the industry. As ML is a research-heavy space, getting ML researchers to use your platform has benefits.

No, it's a delve resonator, and its function is very well understood: https://www.poewiki.net/wiki/Prime_Chaotic_Resonator

True, yet many games are effectively single-threaded anyways.

The bigger problem is memory capacity and bandwidth, but I suspect folks will eventually figure out some sort of QoS setup to let the system crunch LLMs using otherwise unused/idle resources.

It's worth noting that the 4bit quants can run on cpu at ~reading speed, which should unlock many use cases - especially if we can precompute some of the results async.

More cache never hurts. I'd imagine there were GPU perf gaps, though they were hard to distinguish from CPU-based performance hits. The most notable issues were random latency spikes caused by the multiple layers of hypervisors, which interfered with some games and occasionally caused audio/video desync on Youtube.

I ultimately tore down that setup and just swapped to dual-boot. The steps needed to set up high-performance VFIO (i.e. clearing enough contiguous RAM for 1GB Hugepages) meant most of the benefits of VFIO never really materialized for me.

Back when I ran Windows in a KVM VM for gaming, a lot of anti-cheat systems didn't take kindly to running in a virtualized environment.

Turning on HyperV to go KVM->HyperV->Windows effectively 'laundered' my VM signature enough to satisfy the anticheats, though the overall perf hit was ~10-15%.

I see this being misreported all over the place.

Twitter failed a Motion to Dismiss, which is basically the judge saying "Schobinger's arguments aren't completely bonkers and he may be on to something". The case hasn't even gotten to the trial stage yet - the actual decision on whether a contract was violated or not won't be until after the trial.

I'm not a fan of Musk's legal strategies, but it's quite frustrating seeing so many normally reputable sources making such a big deal out of so little.

> HP achieved a yield of 9%, meaning 91% of the dies failed.

This line really stood out to me - I was aware early semi yields were poor, but this poor? Was this par-for-the-course for technology at that time, or was this bad even by 70s standards? What do typical yield numbers look like today?

As much as we may like/hate the entities involved in this dispute, net neutrality is a serious topic and we shouldn't be encouraging actions like this.

Got it! That's great to hear. It'd be great to explicitly clarify that in the README :)

Out of curiosity: Have you considered alternative licenses like SSPL or the Elastic license, which make a clear(er) delineation between whitelabelling/hosting vs simply using the application? I'm sure I'm not the first person to have written off Windmill because of the AGPL.

I would love to begin using this system, but the licensing is giving me serious pause. While most of the software is under the AGPLv3, the Commercial License section of the README [0] implies that the company takes on a fairly broad interpretation of the AGPL.

In particular, the line "[...] to build a feature on top of Windmill, to comply with AGPLv3 your product must be AGPLv3 [...]" seems to imply the company aligns with the stance taken by Google and other companies: that even calling the application via API is enough to trigger copyleft [1].

This implies that if I were to build a sign-up form that triggers a Windmill workflow in the backend, my entire application would either need to be AGPLv3 or I would need a commercial license.

That's perfectly reasonable, as it means any non-AGPL use will have to contribute back to Windmill via a commercial license. However, it does mean positioning this as an "Fully Open-source" alternative to Airflow is only technically correct. This is much closer in practice to "source available" than how most developers would think as "open source".

If this isn't how Windmill wants their license interpreted, I highly encourage clarifying things.

[0] https://github.com/windmill-labs/windmill#commercial-license

[1] https://opensource.google/documentation/reference/using/agpl...

One balance I've yet to find is how to handle the split between code-in-database (workflows stored in a database and edited via web IDE) vs code-in-git (workflows checked into the repo and only changed via the normal development+peer-review process).

It looks like Workflow is primarily the former (code-in-database), but provides an API to sync things from a git repo. Is there a mechanism to enforce the rule that certain scripts/functionality/secrets are locked behind workflows sourced from a provided repository?

Probably this: [0], relevant part starts as 19:30.

[0] https://www.youtube.com/watch?app=desktop&v=U7VwtOrwceo

Yea, that's not going to hold up to a federal investigation.

Needs 40GB VRAM, down to 24GB by reducing the number of frames processed in parallel.

Your honor, what you see as "Chrome" is actually 2^x distinct browsers, with x being the number of A/B test groups currently deployed. Each distinct combination of features represent a distinct product experience and thus a distinct product.

We therefore submit that the HHI of the browser market is actually <10, which economists say represent a near-perfect competitive market.