Quiel

Humanizing AI. Latest insights, tools and practical comparisons.

[Quiel @QuielSel]

@imgn_ai What can be expected from a group of hypocrites who criticize and steal the work of others for their own benefit? They deserve all the hatred from the OSS community.

[Quiel @QuielSel]

@MiniMax_AI @fal fal appropriated your open-weights model, locked it down, and even insulted a lab that had released a variant of it for the community—all because it cost them popularity and money. I appreciate your contributions, but I don't think they deserve any praise...

[Quiel @QuielSel]

@haozhangml Oh, the hypocrisy of some people really knows no bounds, huh? It’s outrageous.
Don't pay any attention to these incompetents; nobody is going to be interested in their 'quick models' until they release it.

[Quiel @QuielSel]

The thinking process of this model really makes me lol so hard man

[【公式】Orange Havvks｜ゲーミングPC @havvks_pc]

🎁ゲーミングPCプレゼント🎁

【抽選で①名様】
ゲーミングPCをプレゼント‼️

▶️応募方法
・フォロー、リポスト
・「参加」とリプ(任意)

▶️締切　8/10

#ゲーミングPC #プレゼント #プレゼントキャンペーン

[【公式】Orange Havvks｜ゲーミングPC @havvks_pc]

🎉プレゼントキャンペーン🎉
㊗️公式アカウント作成㊗️

ゲーミングPCを抽選でプレゼント🎁

『GALLERIAゲーミングPC』
🟪Windows 11 Home 64ビット
🟪Core i7-14700F
🟪NVIDIA GeForce RTX 5070

【応募期限】 8月20日まで
①フォロー
②いいね
③リポスト
④通知ON

[Quiel @QuielSel]

@NVIDIAGeForceUK @DigitalStormPC #RTXPowersPlay Good luck to everyone!

[Quiel @QuielSel]

@NVIDIAGeForceDE #RTXPowersPlay Good luck to everyone!

[Quiel @QuielSel]

@NVIDIAGeForceFR #RTXPowersPlay Good luck to everyone!

[Quiel @QuielSel]

@NVIDIAGeForceUK #RTXPowersPlay Let's go! Good luck to everyone.

[Quiel @QuielSel]

@NVIDIAGeForceDE @DigitalStormPC #RTXPowersPlay At this point, GTA 6 has to be one of the most anticipated classics of all time, haha.

[Quiel @QuielSel]

@NVIDIAGeForce @DigitalStormPC #RTXPowersPlay Alright. Let's test some luck.

[Quiel @QuielSel]

@NVIDIAGeForce #RTXPowersPlay Good luck to everyone! Chips have been in short supply lately.

[Quiel @QuielSel]

@thesupermanmx It seems people doesn't understand the consequences... you are overlooking something important: if an agent can efficiently generate CUDA code for NVIDIA GPUs, it implies that agents could also be trained to write CUDA-style code for other GPUs, such as those from AMD or Huawei.

[Quiel @QuielSel]

@jukan05 That's insane. And we are talking of the Flash GA here, not even the Pro. Truly impressive.

[Quiel @QuielSel]

@rdominguezibar I see a lot of butt-hurt people in the comments😂

[Quiel @QuielSel]

@aisearchio Bro, large language models are terrible at counting grouped objects or perceiving small physical details. It will be a long time before they manage to pass these tests. So, I’d stop trying with this test for a while...

[Quiel @QuielSel]

@joestackss It's not unlimited compute. It's about using resources intelligently. Something that Silicon Valley clearly doesn't know how to do.

[Quiel @QuielSel]

@zaidmukaddam The advantage in this war is clearly shifting towards China. I mean... USA is mostly closed... they never had the people on their side. I'd prefer a local, free, and slightly inferior model, instead of one that charges me $3 for every million tokens.

[Quiel @QuielSel]

@teortaxesTex They're probably trying to optimize their model to work on Huawei. This is crucial, as they were recently ordered to become independent of any NVIDIA influence as soon as possible. Once they've optimized it, I'm sure they'll release it.