Hacker News: yu3zhou4

0/1 Knapsack Problem in PyTorch

yu3zhou4 — Wed, 14 Oct 2026 15:04:33 +0000

Article URL: https://jedrzej.maczan.pl/2025_11_21_dp_knapsack_sliding_hirschberg

Comments URL: https://news.ycombinator.com/item?id=50088767

Points: 51

# Comments: 6

New comment by yu3zhou4 in "AI is creating a 'human premium' for art created by people"

yu3zhou4 — Sun, 11 Oct 2026 12:55:17 +0000

I discuss it a lot with my wife who happens to be a painter.

For years we have perceived my work (swe) as a safe harbor. With AI boom, tides have changed and now I perceive her work as a much, much more future-proof activity

New comment by yu3zhou4 in "Whistle: Speech to Text in 16.9 MB"

yu3zhou4 — Thu, 08 Oct 2026 19:17:07 +0000

I hope you will find a solid solution for your father.

I was researching STT for people with speech disorders two years ago and essentially everything was boiling down to three problems at the end of the day - data scarcity, irregularity of way of speaking and thus constant ambiguity in translation, and individual differences in speech patterns among patients.

New comment by yu3zhou4 in "Kolibri – Tech Report [pdf]"

yu3zhou4 — Sat, 03 Oct 2026 22:20:14 +0000

Much thanks for the explanation and for your help Daniel!

New comment by yu3zhou4 in "Kolibri – Tech Report [pdf]"

yu3zhou4 — Sat, 03 Oct 2026 18:53:45 +0000

It’s not a dupe, it’s a full training and infra report

Kolibri – Tech Report [pdf]

yu3zhou4 — Sat, 03 Oct 2026 17:22:48 +0000

Article URL: https://aleph-alpha.com/downloads/tech-report.pdf

Comments URL: https://news.ycombinator.com/item?id=49946069

Points: 110

# Comments: 5

New comment by yu3zhou4 in ""As a Language Model": Chat Template Switches LLM Self-Referential Voice"

yu3zhou4 — Sun, 27 Sep 2026 11:28:35 +0000

As far as I know we don't know much about metacognition in LLMs, though? Not sure

New comment by yu3zhou4 in ""As a Language Model": Chat Template Switches LLM Self-Referential Voice"

yu3zhou4 — Sun, 27 Sep 2026 11:05:54 +0000

Same! I believe that you could actually train a LoRA on top of a model to get results close to that

New comment by yu3zhou4 in ""As a Language Model": Chat Template Switches LLM Self-Referential Voice"

yu3zhou4 — Sun, 27 Sep 2026 11:05:12 +0000

The strange thing is that the base models (before RLHF) use the "experiential" voice, even though they are not incentivized to do that.

New comment by yu3zhou4 in ""As a Language Model": Chat Template Switches LLM Self-Referential Voice"

yu3zhou4 — Sun, 27 Sep 2026 11:03:36 +0000

Thanks for pointing out, maybe I should be more explicit in the wording - I mean we don't fully know what drives the voice in LLMs. Models that are post trained as instruct models are expected to have the disclaimers, but what about base models (those that are trained on just a lot of text)? How do they talk about themselves? What happens when you strip off the chat template from instruct model's prompt? I hope the rest of the paper makes the questions clearer, but I will try to do better in the abstract next time, as you point out this sentence is kind ambiguous. Thank you!

"As a Language Model": Chat Template Switches LLM Self-Referential Voice

yu3zhou4 — Sun, 27 Sep 2026 10:26:25 +0000

Article URL: https://arxiv.org/abs/2609.25021

Comments URL: https://news.ycombinator.com/item?id=49865343

Points: 103

# Comments: 110

New comment by yu3zhou4 in "Best LLM for every budget, updated daily"

yu3zhou4 — Thu, 24 Sep 2026 16:43:32 +0000

I’d love this but for every hardware

New comment by yu3zhou4 in "Show HN: Drop – a rootless Linux sandbox with gVisor support"

yu3zhou4 — Tue, 22 Sep 2026 14:35:15 +0000

Gratulacje Jan! Looks like something critical to gain adoption these days, security-wise. For others who also wonder how it works, I find this docs page a bit more informative than the landing page https://droprun.sh/docs/sandbox-overview/

What Are Sheaves? (2024)

yu3zhou4 — Sat, 19 Sep 2026 10:31:31 +0000

Article URL: https://www.quantamagazine.org/what-are-sheaves-20240719/

Comments URL: https://news.ycombinator.com/item?id=49765233

Points: 2

# Comments: 0

New comment by yu3zhou4 in "The Secret Life of Circuits"

yu3zhou4 — Sat, 19 Sep 2026 10:27:59 +0000

Michał Zalewski (the author of this book) is goated. His book Silence on the Wire was very inspiring to me and the techniques he described often come to my mind years after reading the book

A CPU Backdoor (2025)

yu3zhou4 — Fri, 18 Sep 2026 21:37:17 +0000

Article URL: https://phrack.org/issues/72/a-cpu-backdoor_md

Comments URL: https://news.ycombinator.com/item?id=49760613

Points: 4

# Comments: 0

New comment by yu3zhou4 in "Ask HN: What are you working on? (September 2026)"

yu3zhou4 — Mon, 14 Sep 2026 07:56:24 +0000

I study stats and probability

New comment by yu3zhou4 in "LLMs and Self-Referentiality"

yu3zhou4 — Thu, 03 Sep 2026 10:40:03 +0000

I found that in LLMs the self-reference defined as "referring in generated text to itself" (so a bit different than in Scott's blog) comes largely from the chat template

https://openreview.net/pdf?id=3O2A23MhNi

Offpunk Manifesto

yu3zhou4 — Wed, 02 Sep 2026 18:33:05 +0000

Article URL: https://blog.ayom.media/ideiasdechirico/offpunk-manifesto

Comments URL: https://news.ycombinator.com/item?id=49540430

Points: 3

# Comments: 0

New comment by yu3zhou4 in "PageRank explained"

yu3zhou4 — Wed, 26 Aug 2026 17:51:35 +0000

Also PageRank as a Markov chain

[0] https://math.libretexts.org/Bookshelves/Linear_Algebra/Under...