Carsen Klock

> No offense intended. It wasn't so much a direct reply as more a warning for future readers that anything on a different chain is not something official, and thus...

> Links redacted, don't click them. Craig Wright is a delusional liar and fraud. > > Get that trash out of here. Craig Wright is a lieing sack of shit...

Tested this slightly, seems to work mostly via running multiple instances of ollama run, did not try via /api/generate endpoint though. One thing I noticed, which may be unrelated, is...

> I'm wondering - why not some kind of queue mechanism? I mean, sometimes response speed is not that crucial and the server's resources are limit - can't requests be...

> @metaspartan > > > Efficiency, Responsiveness, Scalability; it makes Ollama scalable having concurrency vs. having to queue up sequential responses (which it does now). In the future you will...

> > Yes, but then 4 runners could only handle 4 requests at a time (in queue), with this update, it would enable you to send >4 requests and it...

> > They added the new flags back in September: [ggerganov/llama.cpp#3228](https://github.com/ggerganov/llama.cpp/pull/3228) > > I still can't see any kind of queue mechanism - quite the opposite, it feels like it's...

> > Because that is what this is about concurrency/parallelism with ollama. If you want to use it with regular queues just use the default settings or the current release?...

I get this same issue, well actually running the Linux .jar on Mac OS X 10.11.5 El Captain. The OS X .app wouldnt even run (I have Java 8 installed).......

Running `bun install` at all on Windows, results in crash for any command like install or add, bun --version works and bun --help ``` bun install v1.1.0 (5903a614) uh-oh: reached...