Tostino

They'd probably claim "well, it's not an A/B test then, because we are pre selecting for a specific user type who signs up for this".

I don't think that's an excuse, but it is likely the reasoning.

You really don't need an entire new inference engine for this. I modified vllm to work just fine for my decision model experiments.

It properly uses the single shared context and only recomputes the questions, or if the questions are longer, I let the questions go first and then the context that changes often can go after, and then the answer menu tokens... I got roughly the same speed increase with the right workloads that you're saying though.

I hope nothing but the worst for Oracle as a corporation...though nothing against the technical talent there.

If not a windows laptop, a generic GitHub actions runner on whatever hardware they give you.

In this economy?!

I jest...but I'm disgusted by the price gouging going on. Glad my hardware is relatively recent and I had mostly maxed out already when I bought it a few years back. The same setup today would be closer to $7k with equivalent parts.

[flagged]

For their behind closed door conduct? Yes... act on that information how you will.

For getting hacked? No (other than judging the technical team competency).

Same applies here.

Do remember though that the RNC was also hacked at the same time as the DNC, but only one was released right before the election.

I already eat just poultry and some fish...GP's message is still grating for me. I know i'm not being ideologically consistent by continuing to eat any meat, but it is what it is. I can live with that dissonance.

But you will not bring people over to your side with their approach.

The person you originally replied to.

I'm saying they aren't regularly working their horses for hours and hours in that type of heat, day in day out. I'm not saying people don't ride them at all.

But it does seriously take a good bit of hosing them down to lower their internal temp to a safe point after being worked hard. You can't just get them wet and call it good, you need to sit there and keep it flowing over them.

Some of them know the exact danger, and want it to happen because they assume they will be part of the in-group.

Just get out of here and keep watching more slop history on YouTube.

Tik, right?

Anyone have the full article?

Explain to me where in the Constitution it has a carve out for creating an incentive for a private company to gather and collate that data, and then using tax dollars to pay for it.

Uh, maybe it's legal for private citizens to do, but it very much is not constitutional for the government to be doing it.

Gosh darn it, it sure is profitable though!

That can be part of signing the contract for these tools if any thought to security was put in place

Maybe I'm misreading their point...but I'd think that OP was being intellectually consistent and saying that the drug laws should be enforced as written for everyone, or repealed and replaced by more sane laws... because of you were to enforce them as written, you'd have 20-30% (wild ass guess) of the country serving prison time.

And then have to hose them off for an hour so they don't get heat exhaustion. I have a barn full of horses I grew up with in FL. You're misinformed.

Ollama is generally just not great software. Use something else and you likely won't have that issue.

As soon as you scale out to multiple machines, you're already thinking about non-uniform access. If you just push that down one layer further, I don't see all that much of a difference if you're already having to do the work in your software to deal with it.

You don't have a choice when cities, counties, or other government officials decide to make announcements on there.

If a lab doesn't know how to put a new output head on a transformer, they shouldn't be considered a lab.

Why would a Qwen 1.7b be too slow? You should get well under 200ms there.

You better not be someone who ever talks about caring about the Constitution.

How else could you possibly code a GUI?

What they should do, is inject a system prompt at that point telling the model to get out of there. It is the entire reason they train the different levels into the template.

What specific border are you talking about?

Were they charged and convicted with a crime? Maybe they should have been...but we'll never know now.

The available API shapes have limited so much over the years. It's really hard to come up with a new one and have it used, so everyone just tries to fit their work into the existing APIs (chat/completions).

Good job to them for putting out something that does seem quite nice to use, and will likely get a bit of wider traction.