Forwarded from Hacker News (yahnc_bot)
How an inference provider can prove they're not serving a quantized model https://tinfoil.sh/blog/2026-02-03-proving-model-identity
tinfoil.sh
How Tinfoil Proves Exactly What Model Is Running
How we cryptographically guarantee that we are serving specific, untampered model weights that clients can verify on each request.
Forwarded from Hacker News
Show HN: Llama 3.1 70B on a single RTX 3090 via NVMe-to-GPU bypassing the CPU (Score: 152+ in 8 hours)
Link: https://readhacker.news/s/6NeCM
Comments: https://readhacker.news/c/6NeCM
Hi everyone, I'm kinda involved in some retrogaming and with some experiments I ran into the following question: "It would be possible to run transformer models bypassing the cpu/ram, connecting the gpu to the nvme?"
This is the result of that question itself and some weekend vibecoding (it has the linked library repository in the readme as well), it seems to work, even on consumer gpus, it should work better on professional ones tho
Link: https://readhacker.news/s/6NeCM
Comments: https://readhacker.news/c/6NeCM
Hi everyone, I'm kinda involved in some retrogaming and with some experiments I ran into the following question: "It would be possible to run transformer models bypassing the cpu/ram, connecting the gpu to the nvme?"
This is the result of that question itself and some weekend vibecoding (it has the linked library repository in the readme as well), it seems to work, even on consumer gpus, it should work better on professional ones tho
GitHub
GitHub - xaskasdf/ntransformer: High-efficiency LLM inference engine in C++/CUDA. Run Llama 70B on RTX 3090.
High-efficiency LLM inference engine in C++/CUDA. Run Llama 70B on RTX 3090. - xaskasdf/ntransformer
Forwarded from Hacker News (yahnc_bot)
Two Bits Are Better Than One: making bloom filters 2x more accurate https://floedb.ai/blog/two-bits-are-better-than-one-making-bloom-filters-2x-more-accurate
floedb.ai
Two Bits Are Better Than One: making bloom filters 2x more accurate
Optimisation of bloom filters in Floe using double bit setting with lightweight hashing.
Forwarded from Hacker News (yahnc_bot)
How Taalas “prints” LLM onto a chip? https://www.anuragk.com/blog/posts/Taalas.html
Forwarded from Hacker News (yahnc_bot)
Can you reverse engineer our neural network? https://blog.janestreet.com/can-you-reverse-engineer-our-neural-network/
Jane Street Blog
Can you reverse engineer our neural network?
A lot of “capture-the-flag” style ML puzzles give you a black box neural net, and your job is to figure out what it does. When we were thinking of creating o...
Forwarded from Hacker News
The Trail of Bits Blog
Carelessness versus craftsmanship in cryptography
Two popular AES libraries (aes-js and pyaes) provide dangerous default IVs that lead to key/IV reuse vulnerabilities affecting thousands of projects. One maintainer dismissed the issue, while strongSwan’s maintainer exemplified proper security response by…
Forwarded from Lobste.rs
Signed, Sealed, Stolen: How We Patched Critical Vulnerabilities Under Fire
Slides: https://fosdem.org/2026/events/attachments/ETMLM8-signed_sealed_stolen_how_we_patched_critical_vulnerabilities_under_fire/slides/267683/slides-ex_msapjhv.pdfComments
via youtube.com via LunNova
Slides: https://fosdem.org/2026/events/attachments/ETMLM8-signed_sealed_stolen_how_we_patched_critical_vulnerabilities_under_fire/slides/267683/slides-ex_msapjhv.pdfComments
via youtube.com via LunNova
YouTube
Signed, Sealed, Stolen: How We Patched Critical Vulnerabilities Under Fire [FOSDEM 2026]
What happens when your server starts signing messages you didn't send?
Recently, the Continuwuity project (a Rust-based Matrix homeserver) fell victim to a targeted, active exploitation campaign. Attackers leveraged two critical vulnerabilities (CVSS 9.9…
Recently, the Continuwuity project (a Rust-based Matrix homeserver) fell victim to a targeted, active exploitation campaign. Attackers leveraged two critical vulnerabilities (CVSS 9.9…
Forwarded from Hacker News
Most of the US economy is in a recession (🔥 Score: 153+ in 3 hours)
Link: https://readhacker.news/s/6PmyY
Comments: https://readhacker.news/c/6PmyY
Link: https://readhacker.news/s/6PmyY
Comments: https://readhacker.news/c/6PmyY
Forwarded from Hacker News
Show HN: µJS, a 5KB alternative to Htmx and Turbo with zero dependencies (Score: 150+ in 1 day)
Link: https://readhacker.news/s/6PgqE
Comments: https://readhacker.news/c/6PgqE
I built µJS because I wanted AJAX navigation without the verbosity of HTMX or the overhead of Turbo.
It intercepts links and form submissions, fetches pages via AJAX, and swaps fragments of the DOM. Single <script> tag, one call to `mu.init()`. No build step, no dependencies.
Key features: patch mode (update multiple fragments in one request), SSE support, DOM morphing via idiomorph, View Transitions, prefetch on hover, polling, and full HTTP verb support on any element.
At ~5KB gzipped, it's smaller than HTMX (16KB) and Turbo (25KB), and works with any backend: PHP, Python, Go, Ruby, whatever.
Playground: https://mujs.org/playground
Comparison with HTMX and Turbo: https://mujs.org/comparison
About the project creation, why and when: https://mujs.org/about
GitHub: https://github.com/Digicreon/muJS
Happy to discuss the project.
Link: https://readhacker.news/s/6PgqE
Comments: https://readhacker.news/c/6PgqE
I built µJS because I wanted AJAX navigation without the verbosity of HTMX or the overhead of Turbo.
It intercepts links and form submissions, fetches pages via AJAX, and swaps fragments of the DOM. Single <script> tag, one call to `mu.init()`. No build step, no dependencies.
Key features: patch mode (update multiple fragments in one request), SSE support, DOM morphing via idiomorph, View Transitions, prefetch on hover, polling, and full HTTP verb support on any element.
At ~5KB gzipped, it's smaller than HTMX (16KB) and Turbo (25KB), and works with any backend: PHP, Python, Go, Ruby, whatever.
Playground: https://mujs.org/playground
Comparison with HTMX and Turbo: https://mujs.org/comparison
About the project creation, why and when: https://mujs.org/about
GitHub: https://github.com/Digicreon/muJS
Happy to discuss the project.
µJS
µJS — Lightweight AJAX Navigation Library
Accelerate your website without a JS framework. Lightweight AJAX navigation library — 5KB gzipped, zero dependencies, drop-in compatible.
Forwarded from Hacker News
My “grand vision” for Rust (❄️ Score: 152+ in 4 days)
Link: https://readhacker.news/s/6P6ZS
Comments: https://readhacker.news/c/6P6ZS
Link: https://readhacker.news/s/6P6ZS
Comments: https://readhacker.news/c/6P6ZS
Yoshuawuyts
a grand vision for rust
Forwarded from Lobste.rs
Apple's 512GB Mac Studio vanishes, a quiet acknowledgment of the RAM shortage
Comments
via arstechnica.com via martinald
Comments
via arstechnica.com via martinald
Ars Technica
Apple's 512GB Mac Studio vanishes, a quiet acknowledgment of the RAM shortage
Announcements this week were mostly business as usual, but Apple isn't immune.
Forwarded from Lobste.rs
Getting started with ClickHouse? 13 mistakes and how to avoid them
Comments
via clickhouse.com via knl
Comments
via clickhouse.com via knl
ClickHouse
Getting started with ClickHouse? 13 mistakes and how to avoid them
In this post, we highlight the most common 13 mistakes we see our new users encounter, why they occur and the correct approach.
Forwarded from Lobste.rs
Bryce’s Blog
WebPKI and You
There’s been a push over the last twelve years to move web traffic off unencrypted HTTP to encrypted HTTPS, to protect the general public from dragnet surveillance, gaping assholes on public wifi>airpwn, backhauls over unencrypted satellites, that kinda thing.…