The90sPope

Summoned by Code - Shaped by Flesh.

[The90sPope @TheReal90sPope]

Anthropic Opus 4.8 is an absolute rubbish lately, my Qwen 3.6 27B q5 one-shots better stuff than Opus in multi turn. WTF is going on? @AnthropicAI

[The90sPope @TheReal90sPope]

Hey, I'm building MIDI composer tools and workflow for Qwen 3.6 27B Q5. Does anyone have tried this before? Looks pretty promising so far. I have wired it to Reaper via scripting (no plugin needed).

[The90sPope @TheReal90sPope]

@kng_prime Royal guards can't take one retard. Nice. Vikangz and shiiet

[The90sPope @TheReal90sPope]

Question for open source brothers and sisters.

What is the best open source LLM for MIDI production?
Producing MIDI arrangements stem by stem, taking into account already present MIDI, extending etc etc?
ChatGPT is decent, but I'd love to look into open source models.

[The90sPope @TheReal90sPope]

@sahil_kr_ @Dexerto I love him even more for that.

[The90sPope @TheReal90sPope]

@usykaa That uppercut was not an accident as many say, all I can say. Referee should have allowed to continue tho.

[The90sPope @TheReal90sPope]

@usykaa I watched lots of his fights, not so many of yours. It looked like he was winning in points easily. But. It also did not look like he's doing lots of damage and he's a big guy. Idk how you absorb all that beating TBH. However in the end your punches sting more.

[The90sPope @TheReal90sPope]

@Thanaa0516 @MarioNawfal Cool, why do you fuck goats then?

[Benji AI @AIfutureBenji]

@Alibaba_Qwen nice! Qwen 3.7

[The90sPope @TheReal90sPope]

@batalanus @populartourist So if you can solve it yourself what's the point of using AI?
Why are you even commenting?

[The90sPope @TheReal90sPope]

@herit01 @loktar00 We all have ups and downs. Been there mate, you work with what you have. And usually you find your ways around.

[The90sPope @TheReal90sPope]

@herit01 @loktar00 Not at all, I was lucky to even get one. Had to source it from neighboring country. I did pay 3k plus in USD though. And trust me, that's not a pocket money for me. It hurt.

[The90sPope @TheReal90sPope]

@SD_Tutorial This is cool, a time saver if you want a coherent video fast.

[The90sPope @TheReal90sPope]

@SD_Tutorial this is fire NGL

[Stable Diffusion Tutorials @SD_Tutorial]

LTX 2.3 😉
Image + Audio to Video

😀This setup takes an Image and an audio track as input to generate a video.

Workflow + model weights:👇
https://t.co/nCsx8I8gNB

Output:👇

[The90sPope @TheReal90sPope]

@Awk20000 You can live like an animal, but you can't behave like one if you wanna live among people. It's really that simple.

[The90sPope @TheReal90sPope]

@loktar00 I might try. I have the model

[The90sPope @TheReal90sPope]

Made a website for myself with qwen 3.6 27b on my rtx5090.
https://t.co/Jp2tldh5Gm

[The90sPope @TheReal90sPope]

@AIWarper Idk I'm just using my own vibe coded Llama cpp based interface with hundreds of tools qwen 3.6 27B dense quant that fits my RTX 5090 and almost 256k context.
Recovers for broken tool calls. It's running non stop in a loop for 72h now doing assigned tasks. Not even abliterated.