futurepr0n

[futurepr0n @futurepr0n]

@luisjgomez Plus it’s hosted by two bitches

[futurepr0n @futurepr0n]

@SebAaltonen So then I should probably stop trying to build Soldier of Fortune 2, in Rust, shouldn’t I?

[futurepr0n @futurepr0n]

New @timbutterly special from @gasdigital is absolutely must watch 🤣🤣 might even have my wife watch it for the second view. But will it reveal too much truth? https://t.co/QicIjw0QQw

[futurepr0n @futurepr0n]

@ptr_to_joel Haha damn. What is our competent alternative? Hermes? Pi code?

[futurepr0n @futurepr0n]

@quimedesu @faizrazz Thanks - how can someone increase Token per second when running shared ram/vram? Is there any thing I can do? Speed isn’t critical, but hours between prompts would make it impractical for some things, still usable tho for fun

[futurepr0n @futurepr0n]

@mikerainey82 Grow out a goatee, start rocking a singlet or tank top on the regular and just fully commit. You can do this…

[futurepr0n @futurepr0n]

@CrackAmicoRAP @Buckingham_Show Ryan Gosling?!?!

[futurepr0n @futurepr0n]

@richard_froste Did you work on Soldier of Fortune series at all?

[futurepr0n @futurepr0n]

@mikerainey82 @CrackAmicoRAP Trying not to smoke weed and you want to hit me with this on day fucking one…

[futurepr0n @futurepr0n]

@mikerainey82 @CrackAmicoRAP Support our bros ❤️🔥

[futurepr0n @futurepr0n]

@mfranz_on Cries in 3080 TI

[futurepr0n @futurepr0n]

@quimedesu @faizrazz Will try both suggestions thanks. Running with opencode but it could be that which is not effective with this model. Will try pi and Hermes 🙏🤞

[futurepr0n @futurepr0n]

@faizrazz @quimedesu Thanks I have downloaded both but only tested the iq2_xs and it was underwhelming (but hopeful and promising), I am spoiled by my subs to codex and Claude at the moment and trying to pivot to home run LLMs but my gpu power is my bottleneck. 128gb ram tho, but 1 tps is too slow

[futurepr0n @futurepr0n]

@timbutterly

[futurepr0n @futurepr0n]

@faizrazz @quimedesu I’m pretty much finding it unusable with opencode and 12gb gpu. Could be a skill issue but prompts basically fail to recall the the previous (ie changed our port and next prompt rewrote with old port over and over). Nothing built has been good 😭

[futurepr0n @futurepr0n]

@sudoingX I have tried using it and it’s been frustratingly useless compared to Claude or Codex. Forgets things immediately for the next prompt, constant errors in the deliverable. How do you avoid it? I’m using opencode and lmstudio to run it

[futurepr0n @futurepr0n]

@timbutterly This pod is so easy to root for!

[futurepr0n @futurepr0n]

2 session on Max, five hour blocks using “Ultracode” completely burned through without a result to test yet. Feeling more like an ultrascam. What gives @AnthropicAI @claudeai - I will never use higher than low effort again.

[futurepr0n @futurepr0n]

@poiitidis Wizard

[futurepr0n @futurepr0n]

@falco_girgis @thesegaguru Incredible stuff!