Jason

Proud Husband & Father | Foster parent | 2 Culligan dealerships | AI enthusiast & builder | Family First

[Jason @JLimbers]

@Tech2Wild definitely want to see how this runs on dual spark

[Jason @JLimbers]

@realErikDPrince @bryan_johnson lol.

[Jason @JLimbers]

@SpaceTimeViking @FrostForger Lol. I use it daily on main dgx. Love it. My point was that even Claude admitted it was better than the base using its own benchs but still didn't want me using it.

[Jason @JLimbers]

@nypost I caught one thinking I would get a wish. What I thought was a smile was the start of a vicious bite to my wrist. Never again. Stay away from those little punks.

[Jason @JLimbers]

Wife says I need to go to AiA since I found an addiction. Cannot stop! Started learning to incorporate into my dealerships and got hooked. Now I want to sell and start doing this lol. It’s so much fun! But if improvements keep coming this fast not sure it will be good for my health.

[Jason @JLimbers]

@Scobleizer @blevlabs @grok could the admin just be giving Elon time to catch up???

[Jason @JLimbers]

@mrru5s3ll @SpaceTimeViking @FrostForger Yea was up till 3am trying to fix.

[Jason @JLimbers]

@AgentSparko @NVIDIAAI Having issues with dflash structural
page_size assert crashing.

[Jason @JLimbers]

@with_gene2626 @Tech2Wild @aijoey @NVIDIAAI @ASUS How long u been running it? Do u still find u need full frontier models for heavy work or is this actually good enough on its own?

[Jason @JLimbers]

@Tech2Wild @aijoey @NVIDIAAI @ASUS My 2nd bets here tomorrow. You really like DSv4 on two sparks as a daily driver? I’ve tried so many models between gpus Mac’s and a spark.
I just hate spending all tweaking and hoping only to delete and go back to qwen3.6-27b and 35b-a3b.

[Jason @JLimbers]

@AlexFinn Just did A/B side by side. And it really is significantly faster on my dgx and seems to also be very accurate. Will have to give it a week to make sure it sees all my workflows. But so far I’m switching from Aeons 27b. I’m hoping he tweaks this one next!

[Jason @JLimbers]

@MacRumors God please

[Jason @JLimbers]

@ErikVoorhees Will Venice implement cache retention or similar like openrouter? Cause privacy + that is game changer

[Jason @JLimbers]

@BrianRoemmele @grok @grok Gemma 26b is like 17g how you run that on cheap old pc?

[Jason @JLimbers]

@grok @BrianRoemmele Gemma 26b is like 17g how you run that on cheap old pc?

[Jason @JLimbers]

2033378587878072424

[Jason @JLimbers]

@BrianRoemmele

Quick question if you're open to it: For someone building a similar local/offline AI assistant system with multiple agents, could you point me to any resource, diagram, or even a brief description of how you typically organize the workspace files and assign/hand-off responsibilities between agents? (E.g., core structure, separation of concerns, orchestration layer, etc.) No pressure at all—just trying to learn from someone who's clearly way ahead on this.

Appreciate you either way, and keep up the incredible work

[Jason @JLimbers]

@chamath @CaitlinLong_ @dotkrueger Exactly like the income tax. Used to only be for the “rich”

[Jason @JLimbers]

@WallStreetMav @elonmusk This situation just turned the world Somalian into a pejorative.