c/technology

IDPFlare - A fully configurable Identity Provider for Cloudflare. Deploy a complete OAuth/OIDC server in minutes.

Language models typically reason via explicit chain-of-thought (CoT), generating intermediate steps token-by-token. Latent CoT offers an alternative: it performs multi-step reasoning in the model's hidden states, replacing decoded tokens with continuous representations for greater efficiency. However, existing latent CoT methods underperform explicit CoT beyond 1B parameters, and the gap widens with scale. Looped, or recurrent-depth, Transformers, which reuse their weights to increase computation depth without adding parameters, are a natural fit for latent reasoning. We therefore ask whether looped Transformers can bridge this gap. We answer affirmatively with a simple recipe: a looped padded Transformer that processes K latent blocks in parallel for R iterations, with a cross-entropy loss on each latent position's gold CoT-step token, similar to explicit CoT supervision. We instantiate it as LOTUS (Looped Transformers with parallel supervision on latents). LOTUS is, to our knowledge, the first latent-CoT method to bridge the gap to explicit CoT at the 3B scale, while cutting thought-phase latency by 2.5x-6.9x from compact math expressions to natural language. Projecting LOTUS's post-loop latents through the base LM head recovers the gold reasoning steps and even surfaces alternative valid intermediate steps, evidence that its latent space is interpretable and CoT-aligned. Ablations confirm that both the looped backbone and the parallel supervision on gold CoT tokens are essential. Code is available at https://github.com/yingfan-bot/lotus.

I got tired of using onetimesecret.com so I built my own secret sharing service. It has zero dependencies, only standard Web APIs, and the backend...

Battle-tested Serverless AI Skills, Serverless Agent Skills, and serverless skills for your AI coding agents. Instantly add Google Cloud Run, AWS Lambda, Vercel edge, and Microsoft Azure Function skills to your project.

Live Claude Code usage in your menu bar, a multi-session cockpit, and an AI that trims your token waste — a precise instrument that never becomes the engine. 100% local, macOS 14+.

Beautifully designed, copy-paste UI components for AI agent conversations: thinking states, tool calls, streaming text, citations, tables and more. Ships in React, Vue and Svelte with plain CSS and no Tailwind.

We reverse all of the core design principles behind the Macintosh human interface guidelines to arrive at the characteristics of the Internet desktop.

Chat with every big AI model in one native Mac window. Hosted, your own keys, or fully local and offline.

Skip to main content Meet Metal-Organic Frameworks, Chemistry’s New Miracle Materials (photo: California Magazine) September 20, 2018By Krissy Eliot | California MagazineMetal–organic frameworks (MOFs) are a revolutionary new class of crystalline solids that can be designed to trap myriad kinds of matter, including greenhouse gases, or to be used as nanosized drug carriers. They can also pull water from desert air. A chief characteristic of MOFs, which some have been called “miracle materials,” is how much empty space they comprise. Often described in the scientific literature as “ultra highly porous,” the sponge-like nano-structures have the greatest internal surface area of any known material. Unfolded and spread flat, a single gram of MOF could cover a football field. MOFs, as well as other novel chemical structures, including covalent organic frameworks (COFs) and zeolitic imidazolate frameworks (ZIFs), have sprung from a field called reticular chemistry—a name coined by its pioneer, Berkeley chemist Omar Yaghi, who has appointments at both Cal and the Lawrence Berkeley Lab. Professor Omar Yaghi (photo by Marcus Hanschen, UC Berkeley) Professor Yaghi and his Berkeley team made headlines recently for creating an MOF that can harvest water from extremely dry air. The MOFs trap water molecules overnight and release them the next morning as water vapor, which is then condensed as the drinkable stuff. Each pound of MOF powder collects about 1.3 liters of water every 12 hours. Better still, the system can operate on solar energy alone and the MOFs are reusable. The process can repeat night after night. According to Yaghi, the technology is also eminently scalable. “You can use a kilogram [of MOF] or a hundred kilograms if you have a village to water. You can use a thousand kilograms! That would be sitting out there … harvesting water.” Yaghi and his team conceived of the water-harvesting MOF in 2014, when they combined zirconium metal and adipic acid in a structure that bonded to H 2 O. They then joined forces with engineers at MIT who designed the harvester apparatus. Currently, Yaghi and his team are working with companies to commercialize the technology and are eager to get it to people in regions where water is an increasingly scarce and embattled resource. Yaghi pioneered reticular chemistry 20 years ago when he started combining metals with organic molecules in an attempt to better store gasses and liquids. Scientists had tried to synthesize similar chemical structures for decades, Yaghi explained, but they had almost always collapsed. He and his team at Arizona State University succeeded by creating larger clusters that were architecturally robust. Since then the field has grown exponentially. At present, there are some 20,000 different kinds of MOFs and counting, and their uses seem innumerable. “We created chemical structures that have never been made before in numbers and diversity that have never been seen by any class of materials,” Yaghi says. “And now they are fanning out into different applications in energy and environment.” Carbon capture is a prime example. MOFs can be used to mine CO 2 from gasses inside coal-fired power plants before they reach the atmosphere, helping to slow global warming. They can also be used to deliver pesticides more safely and effectively. And medical researchers are using MOFs, which are biodegradable, to carry chemotherapy drugs directly to cancer cells. The next step for MOFs, Yaghi told Berkeley News last year, is to make them more prevalent and to inform people about their potential. “[There] are tremendous problems that are being investigated using MOFs worldwide,” Yaghi said. “What remains is really the widescale deployment of these to actually serve society.” Yaghi’s work has garnered numerous awards, including the Wolf Prize in Chemistry (2018), the Albert Einstein World Award of Science (2017), and the King Faisal International Prize (2015). He’s also in the conversation for a Nobel. It’s all a far cry from his early ambitions. “When I first started in the [field of chemistry],” Yaghi said, “I didn’t go in it to solve society’s problems. I wanted to be left alone, actually.” Krissy Eliot was a little disappointed that the Metal–Organic Framework wasn’t a music festival. From the "Fall 2018 Culture Shift" issue of California Magazine. Topics College of Chemistry topic page MOFs / COFs topic page Recent Stories June 26, 2026 Jun 26 CBE researcher named 2026 Jane Coffin Childs Fellow June 23, 2026 Jun 23 Philomathia Prize advances novel work on the physics of bioelectricity June 17, 2026 Jun 17 Richmond Sarpong honoured with Royal Society of Chemistry Prize News Archive

In an effort to reduce wasteful spending, the government is piloting a program that uses AI for insurance-coverage decisions.

This issue of CCSI Investment Perspectives was contributed by Aniruddha Pravin Jaydeokar. He is a Researcher at the Columbia Center on Sustainable Investment with a focus on leveraging innovative financial structures and instruments to drive systems-level change in mining and energy value chains, as well as in the decarbonization of hard-to-abate sectors. He has done prior research in geothermal energy across Southeast Asia and North America.

We ran one real incident through eleven LLMs. The reasoning turned out to be the easy part; the real challenge is the harness that feeds the model the right context.

SQLite allows NUL characters (ASCII 0x00, Unicode \u0000) in the middle of string values stored in the database. However, the use of NUL within strings can lead to surprising behaviors: The length() SQL function only counts characters up to and excluding the first NUL. The quote() SQL function only shows characters up to and excluding the first NUL. The .dump command in the CLI omits the first NUL character and all subsequent text in the SQL output that it generates. In fact, the CLI omits everything past the first NUL character in all contexts. The use of NUL characters in SQL text strings is not recommended. The SELECT statement above shows output of: (Through this document, we assume that the CLI has ".mode quote" set.) But if you run: Then no rows are returned. SQLite knows that the t1.b column actually holds a 7-character string, and the 7-character string 'abc'||char(0)||'xyz' is not equal to the 3-character string 'abc', and so no rows are returned. But a user might be easily confused by this because the CLI output seems to show that the string has only 3 characters. This seems like a bug. But it is how SQLite works. If you CAST a string into a BLOB, then the entire length of the string is shown. For example: In the BLOB output, you can clearly see the NUL character as the 4th character in the 7-character string. Another, more automated, way to tell if a string value X contains embedded NUL characters is to use an expression like this: If this expression returns a non-zero value N, then there exists an embedded NUL at the N-th character position. Thus to count the number of rows that contain embedded NUL characters: The following example shows how to remove NUL character, and all text that follows, from a column of a table. So if you have a database file that contains embedded NULs and you would like to remove them, running UPDATE statements similar to the following might help: This page was last updated on 2022-05-23 22:21:54Z

One More Letter is a daily word game where every round adds one letter and rearranges them into a new answer.