Hacker News: Azantys

New comment by Azantys in "Hy3"

Azantys — Fri, 10 Jul 2026 12:41:41 +0000

Useless comment

New comment by Azantys in "Hy3"

Azantys — Fri, 10 Jul 2026 10:21:50 +0000

The wording wasnt very good I ment compared to programming or math the amount of logic and reasoning is small (Research level math hardly compares to writing a book in raw reasoning and logic). And I thing the smaller models have enough "intelligence" to write coherent with logical world building, but only the big models can truly do hard math and programming work

New comment by Azantys in "Hy3"

Azantys — Fri, 10 Jul 2026 10:19:36 +0000

Of course there is logic but its nowhere near the complexity of math or programming

New comment by Azantys in "Hy3"

Azantys — Thu, 09 Jul 2026 20:02:23 +0000

Do people really use 100B+ models for writing? I am no writer but to me it seems like writing is one of the easiest tasks with barely any logic or reasoning and as long as its not longer than a handful of pages I expect even 8B models to perform great.

New comment by Azantys in "The gap between open weights LLMs and closed source LLMs"

Azantys — Fri, 26 Jun 2026 22:23:28 +0000

I think model training is pretty hard to do efficiently on a vastly distributed network. If the model cant fit into the VRAM of the node your performance becomes so bad its useless, so a distributed model could only be properly trained if the size of the model doesnt exceed the majority of the nodes VRAM sizes. Maybe there is a different way of doing training but this would be the only way I can see. And it would still be much worse than just using a big datacenter where everything is fully interconnected. BOINC projects work great because its usually just a lot of small compute and memory required so every old desktop and laptop can contribute. Training a model which can compete and is not tiny requires neither low compute or low memory amount. BOINC tasks take minutes usually or sometimes hours but not weeks or months like training a model from scratch. But something like 7B or lower could maybe be trained like this. Im not sure but I think someone is already working on something like this but I dont remember the name of the project.

New comment by Azantys in "Anthropic updates their terms to verify age or identity"

Azantys — Tue, 23 Jun 2026 20:14:33 +0000

New comment by Azantys in "Ask HN: Will programmers write more efficient code during the memory shortage?"

Azantys — Sat, 20 Jun 2026 09:55:00 +0000

Game studios should choose CryEngine/Decima again over UE5

New comment by Azantys in "AMD silently removes memory encryption from consumer Ryzen CPUs"

Azantys — Thu, 18 Jun 2026 11:23:19 +0000

Just dont upgrade the Mainboard firmware then

New comment by Azantys in "DeepSeek Introduces Vision"

Azantys — Thu, 18 Jun 2026 11:12:10 +0000

Yeah but it wasnt close to Opus etc. Still a good local model when it released

New comment by Azantys in "Anthropic's model naming, extrapolated"

Azantys — Wed, 10 Jun 2026 20:12:54 +0000

Galaxy AI 3.8-Flash-Plus Max (xhigh)

New comment by Azantys in "xAI is looking more like a datacentre REIT than a frontier lab"

Azantys — Mon, 08 Jun 2026 18:13:32 +0000

Isnt that more Perplexitys thing anyways?

New comment by Azantys in "xAI is looking more like a datacentre REIT than a frontier lab"

Azantys — Mon, 08 Jun 2026 18:12:32 +0000

Career and personal advice from LLMs, not sure if thats your best bet

New comment by Azantys in "Computex 2026: Intel Launches Crescent Island GPU with Up to 480GB VRAM"

Azantys — Mon, 01 Jun 2026 20:23:23 +0000

Is LPDDR5X not too slow for inference, atleast compared against HBM?

New comment by Azantys in "I found a seashell in the middle of the desert"

Azantys — Sat, 30 May 2026 20:50:29 +0000

I trust a proper solution (even though I can be certain how accurate it is), which compares to a known dataset much more than just giving it an AI. For identifying current living species it is probably fine but this is something to nice for an AI to be trustable. Also this path is much more fun and you learn sonething along the way!

New comment by Azantys in "SANA-WM, a 2.6B open-source world model for 1-minute 720p video"

Azantys — Sat, 16 May 2026 21:09:06 +0000

The Matrix is imminent

New comment by Azantys in "I bought Friendster for $30k – Here's what I'm doing with it"

Azantys — Mon, 27 Apr 2026 06:17:30 +0000

+1 for an android app

New comment by Azantys in "GAIA – Open-source framework for building AI agents that run on local hardware"

Azantys — Tue, 14 Apr 2026 12:47:51 +0000

Yeah I own an AMD Instinct MI50 and i need to patch all of my applications to work, like PyTorch, bitsandbytes, blender etc, while Nvidia cards from the same generation are still mostly supported. But the better value and hardware are worth it

New comment by Azantys in "Arm AGI CPU"

Azantys — Tue, 24 Mar 2026 23:27:58 +0000

It is for orchastrating inference/creating firecracker instances for agents etc. It does'nt have anything to do with actual AI usage.

New comment by Azantys in "15 years of FP64 segmentation, and why the Blackwell Ultra breaks the pattern"

Azantys — Thu, 19 Feb 2026 06:11:38 +0000

Thats because Radeon VIIs were just AMD Instinct MI50 server gpus which didn't make the cut or were left over.