Try this link: https://en.wikipedia.org/wiki/Special:Random

You can keep clicking it too!

Don’t Touch Gooners St. Louis

They said it did better, every news story for their March test says the opposite.

They are an AI hype company. It’s in their interest to tweak the results and the study to make things look better.

They tested it again back in March and got similar results which, being an AI hype company, they buried in their paper.

Many of the programmers refused to not use AI, probably related to the declined cognitive function that some other studies have shown to occur in AI users. I read one testimonial saying that his head would explode if he couldn’t use AI.

So plenty of experience by that point.

You’d be surprised…

An AI hype company did a fairly ok study with programmers. Some used ai some didn’t and they compared the difference in completion time.

The AI users were 20% slower than the non AI users, but thought they were faster.

So for your scenario, the company would need to hire 20 extra people to make up for the lost productivity.

Co-op with my brothers and dad (who was a scuba trainer) is gonna be a blast.

I have a general unease about the sea and water in general so I’m sure it’ll have some lovely tense moments once we get playing.

I love how the url slug for that article seems to imply that it’s about openai stopping shootings.

Similar rant. In the second grade our teacher (FUCK YOU MRS MURRY) had drawn the orbit of the Earth around the sun and was telling us that because it was elliptical and that’s why we had summer; the Earth was closer to the sun and the sun was warm.

She basically drew an oval on the chalkboard and put the sun smack in the center. It didn’t make any sense to me so I kept asking why there weren’t two summers in a year if an orbit was a year and the earth passed close the sun twice…

It wasn’t until the 3rd or 4th grade when I got a hold of an illustrated astronomy book that showed our titled planet and explained the seasons.

Yup! If you installed the os on it.

So you have one usb with the iso flashed to it and a second to install the os on. Use the first to install to the second.

So a trick for the double drives is to pop in a low profile usb drive and install the os on that. Then you can use the ssd/hdd for other things.

Can’t you just use the ellipse selection and do a trace selection?

I’ve been wondering if putting most of the tube in the tank and then covering up the hole with your thumb would be adequate in pulling up liquid to begin the siphon process.

My previous boss (the CIO) began using AI back in May of last year in a desperate attempt to not get fired. He had spearheaded a shift to a SaaS product that ended up costing us 200x more than our in house solution and was in hot water for that. Additionally a new COO was about to be hired who was previously a CIO and might figure out that he didn’t really know anything.

Since he had basically bullshitted his way into the position I think he turned to a bullshit engine to assist him in saving himself. Ultimately his efforts were futile and he was let go.

The new COO appears to be just as enthralled to AI… She was in my office the other day asking me questions about our MDM, feeding my answers into copilot, and then asking my follow up questions that it gave her.

Beautiful, maybe I should get an mba.

Good response. Love it when peeps own up to their mistskes.

100% agree. At this point I am assuming everything sent through their servers is actively being collected for LLM training.

Unless someone has released something new while I haven’t been paying attention, all the gen AIs are essentially frozen. Your use of them can’t impact the actual weights inside of the model.

If it seems like it’s remember things is because of the actual input of the LLM is larger than the input you will usually give it.

For instance lets say the max input for a particular LLM is 9096 tokens. The first part of that will be instructions from the owners of the LLM to prevent their model from being used for things they don’t like. Lets say the first 2000 tokens. That leaves 7k or so for a conversation that will be ‘remembered’.

Now if someone was really savvy, they’d have the model generate summaries of the conversation and stick them into another chunk of memory, maybe another 2000 tokens worth, that way it will seem to remember more than just the current thread. That would leave you with 5000 tokens to have a running conversation.

Can we add another line for depth as well?

Here’s mine, no judging for how I set the ip

name: jellyfin
services:
    jellyfin:
        container_name: jellyfin.live
        networks:
            mynet:
                ipv4_address: 192.168.5.3
        ports:
            - 8096:8096
        volumes:
            - jellyfin-config:/config
            - jellyfin-cache:/cache
            - type: bind
              source: /mnt
              target: /Media Disk 1
            - type: bind
              source: /mnt
              target: /Media Disk 2
            - type: bind
              source: /mnt
              target: /Media Disk 3
        environment:
            - HEALTHCHECK_URL=192.168.5.3
        restart: unless-stopped
        deploy:
            resources:
                reservations:
                    devices:
                        - driver: nvidia
                          count: all
                          capabilities:
                              - gpu
        image: jellyfin/jellyfin:latest
networks:
    mynet:
        external: true
        name: mynet
volumes:
    jellyfin-config:
        external: true
        name: jellyfin-config
    jellyfin-cache:
        external: true
        name: jellyfin-cache

The section on the GPU looks like yours. I had to do something to get it working, but it’s been forever.

Budget-Chicken-2425

Looks like his account is suspended now.

What will Vinny play now then? :(