Kerosene did not save the sperm whale (2024) (Score: 150+ in 9 hours)
Link: https://readhacker.news/s/6sc59
Comments: https://readhacker.news/c/6sc59
Link: https://readhacker.news/s/6sc59
Comments: https://readhacker.news/c/6sc59
Substack
No, Kerosene did NOT save the Sperm Whale
How we carried on exploiting sperm whales long after the conventional wisdom said we'd stopped. Forgotten Material #1
Trump's Tariffs Wipe Out over $6T on Wall Street in Epic Two-Day Rout (🔥 Score: 189+ in 48 minutes)
Link: https://readhacker.news/s/6sdD9
Comments: https://readhacker.news/c/6sdD9
Link: https://readhacker.news/s/6sdD9
Comments: https://readhacker.news/c/6sdD9
WSJ
Trump’s Tariffs Wipe Out Over $6 Trillion on Wall Street in Epic Two-Day Rout
The selloff capped the worst week since 2020, as fears of a recession mount
The 'Judicial Black Hole' of El Salvador's Prisons Is a Warning for Americans (🔥 Score: 159+ in 2 hours)
Link: https://readhacker.news/s/6sdyu
Comments: https://readhacker.news/c/6sdyu
Link: https://readhacker.news/s/6sdyu
Comments: https://readhacker.news/c/6sdyu
Rolling Stone
The ‘Judicial Black Hole’ of El Salvador’s Prisons Is a Warning for Americans
Donald Trump is cementing a partnership with El Salvador President Nayib Bukele, whose assault on civil rights is being mirrored in the U.S.
Multi-Token Attention (❄️ Score: 150+ in 2 days)
Link: https://readhacker.news/s/6s55J
Comments: https://readhacker.news/c/6s55J
Link: https://readhacker.news/s/6s55J
Comments: https://readhacker.news/c/6s55J
arXiv.org
Multi-Token Attention
Soft attention is a critical mechanism powering LLMs to locate relevant parts within a given context. However, individual attention weights are determined by the similarity of only a single query...
Coolify: Open-source and self-hostable Heroku / Netlify / Vercel alternative (❄️ Score: 151+ in 2 days)
Link: https://readhacker.news/s/6s33E
Comments: https://readhacker.news/c/6s33E
Link: https://readhacker.news/s/6s33E
Comments: https://readhacker.news/c/6s33E
Coolify
An open-source & self-hostable Heroku / Netlify / Vercel alternative.
Learn electricity and electronics fundamentals without taking a formal course (🔥 Score: 151+ in 3 hours)
Link: https://readhacker.news/s/6sdNS
Comments: https://readhacker.news/c/6sdNS
Link: https://readhacker.news/s/6sdNS
Comments: https://readhacker.news/c/6sdNS
Trump's Tariff Formula Makes No Economic Sense. It's Also Based on an Error (🔥 Score: 150+ in 2 hours)
Link: https://readhacker.news/s/6se2p
Comments: https://readhacker.news/c/6se2p
Link: https://readhacker.news/s/6se2p
Comments: https://readhacker.news/c/6se2p
American Enterprise Institute - AEI
President Trump’s Tariff Formula Makes No Economic Sense. It’s Also Based on an Error.
President Trump on Wednesday announced tariffs on practically every foreign country (and some non-countries), ranging from a 10 percent minimum all the way up to 50 percent. He described the tariffs as reciprocal; however, they are nothing of the sort.
Nebula Sans (🔥 Score: 150+ in 3 hours)
Link: https://readhacker.news/s/6segK
Comments: https://readhacker.news/c/6segK
Link: https://readhacker.news/s/6segK
Comments: https://readhacker.news/c/6segK
Nebulasans
Nebula Sans
Nebula Sans is the brand typeface for Nebula, available under the SIL Open Font License.
Recreating Daft Punk's Something About Us (Score: 150+ in 9 hours)
Link: https://readhacker.news/s/6sedC
Comments: https://readhacker.news/c/6sedC
Link: https://readhacker.news/s/6sedC
Comments: https://readhacker.news/c/6sedC
Thoughts and things
Recreating Daft Punk's Something About Us
An in-depth, personal journey recreating Daft Punk’s “Something About Us” in Ableton Live 12—track by track. From synths to talkbox guitars, a nostalgic dive into French Touch culture, sound design, and the joys of music production.
Annotated Unix Magic Poster (Score: 152+ in 16 hours)
Link: https://readhacker.news/s/6sdzL
Comments: https://readhacker.news/c/6sdzL
Link: https://readhacker.news/s/6sdzL
Comments: https://readhacker.news/c/6sdzL
Google announces Sec-Gemini v1 a new experimental cybersecurity model (Score: 150+ in 22 hours)
Link: https://readhacker.news/s/6scRu
Comments: https://readhacker.news/c/6scRu
Link: https://readhacker.news/s/6scRu
Comments: https://readhacker.news/c/6scRu
Google Online Security Blog
Google announces Sec-Gemini v1, a new experimental cybersecurity model
Posted by Elie Burzstein and Marianna Tishchenko, Sec-Gemini team Today, we’re announcing Sec-Gemini v1, a new experimental AI model focused...
What If We Made Advertising Illegal? (🔥 Score: 163+ in 52 minutes)
Link: https://readhacker.news/s/6sfyX
Comments: https://readhacker.news/c/6sfyX
Link: https://readhacker.news/s/6sfyX
Comments: https://readhacker.news/c/6sfyX
Kōdō Simone
What If We Made Advertising Illegal?
What if we banned all advertising? Not regulate it—abolish it. This proposal would transform manipulation machines, and maybe save democracy itself. A thought experiment worth considering.
Llama4 (🔥 Score: 202+ in 36 minutes)
Link: https://readhacker.news/s/6sfEB
Comments: https://readhacker.news/c/6sfEB
Link: https://readhacker.news/s/6sfEB
Comments: https://readhacker.news/c/6sfEB
Llama
Llama 4 is Here | Meta
Leading intelligence. Unrivaled speed and efficiency.
Show HN: I built a word game. My mom thinks it's great. What do you think? (Score: 152+ in 4 hours)
Link: https://readhacker.news/s/6sf6x
Comments: https://readhacker.news/c/6sf6x
Link: https://readhacker.news/s/6sf6x
Comments: https://readhacker.news/c/6sf6x
www.whatsit.today
What's it? - Word Unscrambling Game
Unscramble hidden words in daily grid puzzles based on fun categories!
OpenVertebrate Presents a Database of 13,000 3D Scans of Specimens (Score: 151+ in 17 hours)
Link: https://readhacker.news/s/6sdSF
Comments: https://readhacker.news/c/6sdSF
Link: https://readhacker.news/s/6sdSF
Comments: https://readhacker.news/c/6sdSF
Open Culture
OpenVertebrate Presents a Massive Database of 13,000 3D Scans of Vertebrate Specimens
From The Florida Museum of Natural History comes the openVertebrate project, a new initiative to 'provide free, digital 3D vertebrate anatomy models and data to researchers, educators, students and the public.' Introducing the new project (otherwise known…
Emulating an iPhone in QEMU (Score: 151+ in 9 hours)
Link: https://readhacker.news/s/6seDT
Comments: https://readhacker.news/c/6seDT
Link: https://readhacker.news/s/6seDT
Comments: https://readhacker.news/c/6seDT
Earth's clouds are shrinking, boosting global warming (Score: 150+ in 8 hours)
Link: https://readhacker.news/s/6seL6
Comments: https://readhacker.news/c/6seL6
Link: https://readhacker.news/s/6seL6
Comments: https://readhacker.news/c/6seL6
Science
Earth’s clouds are shrinking, boosting global warming
Narrowing storm bands may be a surprising and dangerous new feedback of climate change
A Vision for WebAssembly Support in Swift (Score: 152+ in 8 hours)
Link: https://readhacker.news/s/6sf36
Comments: https://readhacker.news/c/6sf36
Link: https://readhacker.news/s/6sf36
Comments: https://readhacker.news/c/6sf36
Swift Forums
[Pitch] A Vision for WebAssembly Support in Swift
As WebAssembly support has been developed by the Swift community and significantly improved over the years, I would like to put up a pitch for a vision describing WebAssembly support in Swift. Your feedback would be highly appreciated! Full vision text is…
DeepSeek: Inference-Time Scaling for Generalist Reward Modeling (Score: 150+ in 1 day)
Link: https://readhacker.news/s/6sacg
Comments: https://readhacker.news/c/6sacg
Link: https://readhacker.news/s/6sacg
Comments: https://readhacker.news/c/6sacg
arXiv.org
Inference-Time Scaling for Generalist Reward Modeling
Reinforcement learning (RL) has been widely adopted in post-training for large language models (LLMs) at scale. Recently, the incentivization of reasoning capabilities in LLMs from RL indicates...
Show HN: OCR pipeline for ML training (tables, diagrams, math, multilingual) (Score: 150+ in 19 hours)
Link: https://readhacker.news/s/6secG
Comments: https://readhacker.news/c/6secG
Hi HN,
I’ve been working on an OCR pipeline specifically optimized for machine learning dataset preparation. It’s designed to process complex academic materials — including math formulas, tables, figures, and multilingual text — and output clean, structured formats like JSON and Markdown.
Some features:
• Multi-stage OCR combining DocLayout-YOLO, Google Vision, MathPix, and Gemini Pro Vision
• Extracts and understands diagrams, tables, LaTeX-style math, and multilingual text (Japanese/Korean/English)
• Highly tuned for ML training pipelines, including dataset generation and preprocessing for RAG or fine-tuning tasks
Sample outputs and real exam-based examples are included (EJU Biology, UTokyo Math, etc.)
Would love to hear any feedback or ideas for improvement.
GitHub: https://github.com/ses4255/Versatile-OCR-Program
Link: https://readhacker.news/s/6secG
Comments: https://readhacker.news/c/6secG
Hi HN,
I’ve been working on an OCR pipeline specifically optimized for machine learning dataset preparation. It’s designed to process complex academic materials — including math formulas, tables, figures, and multilingual text — and output clean, structured formats like JSON and Markdown.
Some features:
• Multi-stage OCR combining DocLayout-YOLO, Google Vision, MathPix, and Gemini Pro Vision
• Extracts and understands diagrams, tables, LaTeX-style math, and multilingual text (Japanese/Korean/English)
• Highly tuned for ML training pipelines, including dataset generation and preprocessing for RAG or fine-tuning tasks
Sample outputs and real exam-based examples are included (EJU Biology, UTokyo Math, etc.)
Would love to hear any feedback or ideas for improvement.
GitHub: https://github.com/ses4255/Versatile-OCR-Program
GitHub
GitHub - ses4255/Versatile-OCR-Program: Multi-modal OCR pipeline optimized for ML training (text, figure, math, tables, diagrams)
Multi-modal OCR pipeline optimized for ML training (text, figure, math, tables, diagrams) - ses4255/Versatile-OCR-Program