2 months agodontwatchme____
Bredda how are you 41. I watched your whole career and I’m 33. Time is a madness
2 months agopaquino12
Don’t subscribe to perplexity it’s a real scam, you pay and the subscription is not worth the time it asks you for more money to buy credits, no one explains it to you before this is a huge fraud
2 months agovictoriousmarcus1
🎇 Hey 😊 👏👏👏👏😱 PLS VOTE FOR GRAMMYS Timbaland Collaboration & $100K NOW 1st PLZ LIKE/SHARE :) THANKFUL I WON 2955 AWARDS/GRAMMY NOM/EMMY PARTNER/83 GRAMMYS Collaboration/ 27 BILLIONS VIEWS other project Big Movies TV MUCH LOVE GOD BLESS :) https://orbiiit.com/en/participants/cfede2dc-66a4-4385-98e0-cd5abddb1c19?contestId=f85717be-ba9b-4857-b885-ccbbb9a45757
2 months agoiphone_18prowifi8
@nvidia I strongly recommend training all of NVIDIA’s AI models from scratch, rather than fine-tuning existing models, on NVIDIA’s 2027 Vera Rubin Ultra platform, featuring LPDDR6 PIM SOCAMM2 memory. Only training AI models from scratch—not fine-tuning them—would allow the full, enormous potential of the NVIDIA Vera Rubin Ultra platform with LPDDR6 PIM SOCAMM2 memory to be realized. Fine-tuning would carry forward many limitations and legacy constraints inherited from older platforms such as NVIDIA Volta (2018), NVIDIA Ampere (2020), NVIDIA Hopper (2022), NVIDIA Blackwell (2024), and NVIDIA Blackwell Ultra (2025). I propose bringing this idea directly to Jensen Huang and encouraging him to begin training all of NVIDIA’s AI models from scratch on the 2027 NVIDIA Vera Rubin Ultra platform with LPDDR6 PIM SOCAMM2 memory. Let’s work together to make AI truly great.
More posts
