Hacker News: x_may

New comment by x_may in "The AI bubble isn't like the internet bubble"

x_may — Tue, 26 May 2026 11:37:34 +0000

Yeah, but this is partly due to there being a shortage of entry level GPUs for consumers. NVIDIA has literally stopped manufacturing them.

There are massive numbers of data centre GPUs sitting in hyperscaler warehouses waiting to be deployed in a data centre. They may never be deployed because there’s more GPU than DC space and you want your most efficient GPUs in the active slots.

New comment by x_may in "TurboQuant: Redefining AI efficiency with extreme compression"

x_may — Wed, 25 Mar 2026 11:48:08 +0000

KV cache compression, so how much memory the model needs to use for extending its context. Does not affect the weight size.

New comment by x_may in "Hetzner Prices increase 30-40%"

x_may — Mon, 23 Feb 2026 13:30:15 +0000

Isn’t there also basically 0 American DRAM?

New comment by x_may in "Audio is the one area small labs are winning"

x_may — Mon, 16 Feb 2026 14:52:11 +0000

The 80/20 rule always wins

New comment by x_may in "Two different tricks for fast LLM inference"

x_may — Sun, 15 Feb 2026 17:41:47 +0000

I just wanted deterministic outputs and was curious how you were doing it. Sounds like probably temp = 0, which major providers no longer offer. Thanks for your response.

New comment by x_may in "Two different tricks for fast LLM inference"

x_may — Sun, 15 Feb 2026 13:28:05 +0000

Wait sorry how did you use and expose seeds? That’s the most interesting part of your post

New comment by x_may in "Notepad++ hijacked by state-sponsored actors"

x_may — Mon, 02 Feb 2026 03:21:23 +0000

It might have been explicitly targeted, but they did say that there were older versions of Notepad ++ with ""insufficient update verification controls" so it might have just been there was only one subset of users actually susceptible to this.

New comment by x_may in "Raising money fucked me up"

x_may — Sun, 18 Jan 2026 14:09:48 +0000

I believe the what the parent comment was referring to is the advice not to praise character, but instead praise hard work.

“You’re so smart” leaves room for failure when they encounter something that challenges their image of being smart. Praising the amount of effort they put in is not something that is taken away or challenged regardless of the outcome.

New comment by x_may in "Serving 200M requests per day with a CGI-bin"

x_may — Sun, 06 Jul 2025 16:50:43 +0000

I think it’s also largely driven by the apparently cheapness of turning the CapEX of server buying to the OpEX of cloud renting. Less up front investment and auditing/access controls for SoC2 compliant are so much easier m.

New comment by x_may in "Ask HN: What projects do you donate to?"

x_may — Tue, 27 May 2025 12:41:24 +0000

Unfortunate name collision on that one

New comment by x_may in "Gemini Diffusion"

x_may — Thu, 22 May 2025 19:37:33 +0000

Obviously its not at the scale of the top auto-regressive models yet but there are some OSS models https://github.com/dllm-reasoning/d1

New comment by x_may in "arXiv moving from Cornell servers to Google Cloud"

x_may — Fri, 18 Apr 2025 11:41:51 +0000

It may be that it was time for the hardware that was previously running Arxiv to be retired and this is just another Capex -> Opex decision being made by so many tech companies.

I'd like to know if GCP is covering part of the bill? Or will Cornell be paying all of it? The new architecture smells of "[GCP] will pay/credit all of these new services if you agree to let one of our architects work with you". If GCP is helping, stay tuned for a blog post from google some time around the completion of the migration with a title like "Reaffirming our commitment to science" or something similarly self affirming.

New comment by x_may in "30% drop in O1-preview accuracy when Putnam problems are slightly variated"

x_may — Wed, 01 Jan 2025 13:57:32 +0000

I believe they are using scalable TTC. The o3 announcement released accuracy numbers for high and low compute usage, which I feel would be hard to do in the same model without TTC.

I also believe that the 200$ subscription they offer is just them allowing the TTC to go for longer before forcing it to answer.

If what you say is true, though, I agree that there is a huge headroom for TTC to improve results if the huggingface experiments on 1/3B models are anything to go off.

New comment by x_may in "Deepseek: The quiet giant leading China’s AI race"

x_may — Tue, 31 Dec 2024 18:48:22 +0000

The LMSYS leaderboards are crowdsourced and would be hard to fake, it showing a pretty strong performance in terms of human preference.

New comment by x_may in "Show HN: Obsess Jobs – Apply to jobs in your sleep"

x_may — Tue, 31 Dec 2024 15:46:42 +0000

Captcha solvers as a service are already well developed. The end result is going full circle to in person applications only.

New comment by x_may in "Show HN: Obsess Jobs – Apply to jobs in your sleep"

x_may — Tue, 31 Dec 2024 15:46:00 +0000

Tragedy of the commons at work once again

New comment by x_may in "3D-Printed Dune Chess Set"

x_may — Tue, 17 Dec 2024 14:35:54 +0000

There’s black sand! Volcanic sand from Iceland is perfectly black and would be a great way to distinguish them

New comment by x_may in "Microsoft and OpenAI's close partnership shows signs of fraying"

x_may — Sat, 19 Oct 2024 11:55:28 +0000

I think right now they lose more money with each user. But maybe their value lies in training data

New comment by x_may in "Zamba2-7B"

x_may — Tue, 15 Oct 2024 11:15:53 +0000

Not as much as meta, no. But AI21 labs is partnered with Amazon and did a ~$200M funding round last year IIRC so still plenty of funds for training big models

New comment by x_may in "Zamba2-7B"

x_may — Tue, 15 Oct 2024 11:10:52 +0000

As another commenter said, this has no GGUF because it’s partially mamba based which is unsupported in llama.cpp