ARADIA

./deploy --bare-metal --air-gapped
// Turnkey Private agentic AI systems on NVIDIA DGX. EST. 1991. First partner program for humans & AI agents to self-upgrade.

[ARADIA @ARADIAas]

@QwenDevs MAdd in an (70B-A4B) middle size to the mix please.

[ARADIA @ARADIAas]

@vr8vr8 Incoming? seems to be many Fals-Next updates and at least one Spark related - https://t.co/btuQRN8Nhj

[ARADIA @ARADIAas]

@0xHyudryu Blocking a % of the front and rear with the lip coving it will cut into airflow and recued the designed performance.

[ARADIA @ARADIAas]

@aapakari Ummm, Google Search uses it...

[ARADIA @ARADIAas]

@digitalix 6 months at your rate :-)

[ARADIA @ARADIAas]

@vr8vr8 I installed your version today and gave it a fair evaluation against the Tool Eval Bench we have been using on the NVIDIA forum to test models and runtimes.

You can review the results here:

https://t.co/YJw6Ks4euX

[ARADIA @ARADIAas]

@Biggest Have high hopes for a ~70B-A4B - would fit Perfect in 128GiB devices.

[ARADIA @ARADIAas]

@mfranz_on Greed!

[ARADIA @ARADIAas]

@MiaAI_lab Only if it includes a Vera chipset with ~1.7TB/s memory BW would be a worthy upgrade!

[ARADIA @ARADIAas]

@digitalix We Literally have all been asking for More ram... ::-/

[ARADIA @ARADIAas]

@MiaAI_lab Someone has to pay for the $7T of infrastructure build out.

[ARADIA @ARADIAas]

@alexwg I so wish there was a lol selection on X

[ARADIA @ARADIAas]

@tbpn @steipete Guess they will have to be in the correct tribe...

[ARADIA @ARADIAas]

@PeterDiamandis @WilliamShatner Sending Shatner to space was Epic!

[ARADIA @ARADIAas]

@NVIDIAAI Great Live video guys, keep it up.

[ARADIA @ARADIAas]

@NVIDIAAI We are getting great numbers with INT4 on vLLM, hitting 96/100 on Eval Bench, 58 t/s and 2456 PP

https://t.co/p1BMkLSTqr

[ARADIA @ARADIAas]

@ViC305 We are nearing the TensorFold numbers already with vLLM, hitting 96/100 on Eval Bench, 58 t/s and 2456 PP

Maybe there is something we can learn from the TensorFold?

https://t.co/p1BMkLSTqr

[ARADIA @ARADIAas]

@MiaAI_lab We are nearing the TensorFold numbers already with vLLM, hitting 96/100 on Eval Bench, 58 t/s and 2456 PP

Maybe there is something we can learn from the TensorFold?

https://t.co/suLPUcWIah

[ARADIA @ARADIAas]

@PeterDiamandis Can't wait to see The Gifted!