Search the archive

DOCMOST

Docmost is an open source collaborative documentation and wiki software similar to Confluence and Notion.

WHAT DOES WORK-LIFE BALANCE EVEN MEAN?

Work-life balance is influenced by intrinsic motivation, opportunity cost, and health.

how generative AI is increasing efficiency and improving accuracy

across core financial functions, including:

SPACEX TENDER OFFER SAID TO VALUE COMPANY AT RECORD $210 BILLION

SpaceX will sell insider shares at $112 each in a tender offer that values the company at close to $210 billion. The company was valued at $180 billion in December. The new valuation is a record for an American private company. It puts SpaceX on par with some of the world's largest publicly traded companies by market capitalization. Terms on the tender offer haven't been finalized - the size of the offer could change depending on interest from both insider sellers and buyers.

APPLE DEVELOPING NEW WAY TO MAKE IPHONE BATTERIES EASIER TO REPLACE

Apple is reportedly investigating 'electrically induced adhesive debonding' technology to make iPhone batteries more replaceable. The technology would allow a battery to be dislodged from its chassis with a small jolt of electricity to the battery. It could debut with at least one of the iPhone 16 models this year.

PIPING UP AT THE GATES OF DAWN

Astronomers have found a galaxy that was already shining with intense starlight when the universe was only 290 million years old. The earliest, most distant galaxy yet found, JADES-GS-z14-0 is one of a string of discoveries enabled by the James Webb Space Telescope that challenge conventional models of how the first stars and galaxies formed. The discovery proves that luminous galaxies were already in place 300 million years after the Big Bang and are more common than what was expected. Galaxy formation models will need to address the existence of such large and luminous galaxies so early in cosmic history.

GERMAN FIRM SYNHELION OPENS ‘WORLD'S 1ST' INDUSTRIAL SOLAR FUEL PLANT

DAWN is a project from Synhelion that aims to remove the dependency on fossil fuels from the transport sector, particularly in aviation. It uses a 65-foot-high solar tower surrounded by a mirror field that concentrates solar rays to produce synthetic fuels using solar heat. The tower has an energy storage system that allows it to generate solar fuel even when there's no sunlight. DAWN is scheduled to begin production in the latter half of the year. Synhelion says DAWN will produce several thousand liters of fuel per year.

solution brief

CORCEL

Corcel is a collection of PHP classes that provides a fluent interface for connecting and getting data directly from a WordPress database. It allows WordPress to be used as a backend. Corcel can be used with any PHP project that uses Composer.

DORKLY FEATURE FLAGS

Dorkly is a free feature flag system for LaunchDarkly's open source SDKs. It allows developers to implement feature flagging consistently across dozens of languages. Dorkly was designed to fit into existing workflows. It aims to be a simple feature flagging system without the cognitive load of yet another tool.

LEARNING FROM POPUP VILLAGES

Some popup villages have seen hundreds of people from the internet gather for multiple months at a time. These gatherings are long enough to build relationships but short enough to not require permanent relocation. People can join these communities for however long they want, creating an opportunity for a demand-first approach to prototyping permanent places. This essay looks at successful popup villages, why they work, and where they could go next.

OPENAI'S NEW “CRITICGPT” MODEL IS TRAINED TO CRITICIZE GPT-4 OUTPUTS

CriticGPT is a new AI model from OpenAI designed to identify mistakes in code generated by ChatGPT. Based on the GPT-4 family of large language models, it analyzes code and points out potential errors, making it easier for humans to spot mistakes that might otherwise go unnoticed. CriticGPT was trained on a large number of inputs containing deliberately inserted mistakes. Its capabilities extend beyond just code review - it is able to identify errors made by ChatGPT in some other tasks as well. OpenAI plans to integrate CriticGPT-like models into its RLHF labeling pipeline to provide its trainers with AI assistance.

workspaces. Watch the video (Youtube)

TIKTOK TO CHALLENGE AMAZON PRIME DAY WITH ITS OWN SALES EVENT IN JULY

TikTok Shop is holding a 'Deals For You Days' sales event in the US starting on July 9.

ON CLAUDE 3.5 SONNET

Claude 3.5 Sonnet is currently the best non-tiny large language model.

MAKER OF RSTUDIO LAUNCHES NEW R AND PYTHON IDE

The Positron IDE, based on VS Code and designed specifically for R and Python, is available in public beta for macOS, Windows, and Linux.

META'S LLM COMPILER IS THE LATEST AI BREAKTHROUGH TO CHANGE THE WAY WE CODE

The Meta Large Language Model Compiler is a suite of models designed to emulate the compiler, predict optimal passes for code size, and disassemble code.

HOW TO THINK ABOUT CREATING A DATASET FOR LLM FINETUNING EVALUATION

This article summarizes the kinds of evaluations needed for a structured data generation task.

I'VE GONE BACK TO MY THINKING ABOUT THE LONG-TERM CONSEQUENCES OF ALTMAN'S COUP

OpenAI's magic seems to have worn off - the company doesn't have that much of a moat without its people.

algolia.com

SNAP LENSE STUDIO 5.0

Snap has released the GenAI suite in Lens Studio 5.0 - this is a great advancement and is massively useful for developing AR applications.

INSTAGRAM LAUNCHING AN AI STUDIO

Instagram's new "AI Studio" lets creators develop AI chatbot versions of themselves. It is currently rolling out as an early test in the US.

APPLE, MICROSOFT SHRINK AI MODELS TO IMPROVE THEM

LLM COMPILER

Meta has released two language models that can compile code to assembly and decompile to LLVM IR. They were trained on 546 Billion tokens of high-quality data then further instruction tuned. They are able to achieve 77% of optimized assembly performance and have 45% round trip disassembly performance.

CRITICGPT

OpenAI has trained a model based on GPT-4 that can catch GPT-4 mistakes and help humans spot mistakes during the RLHF process.

A BENCHMARK FOR ACTIVE LEARNING PIPELINES

ALPBench is a tool designed to standardize benchmarks for active learning query strategies.

and securely. Get started or book a demo

Need to deploy AI /ML at the edge? Deliver fast AI applications

with high throughput and ultra-low latency anywhere in the world with Gcore [12]. Automatically adjust resources based on demand (only pay for what you use), deploy and scale effortlessly, process data locally

FAL GIGAGAN UPSCALER

AuraSR is a super resolution image upscaler recently open sourced by FAL. It requires a single forward pass and can upscale by 4x even over repeated applications. AuraSR works well on generated images.

3D SEGMENTATION WITH INTERACTIVE GUIDANCE

To meet the growing demand for detailed 3D data, researchers have introduced Point-SAM, a transformer-based model for 3D segmentation.

GENERATIVE INFORMATION RETRIEVAL

This survey delves into generative information retrieval (GenIR), a fresh take on finding information that moves away from traditional search methods to ones that create answers on the fly.

ARE TECH-ENABLED VERTICAL ROLL-UPS THE FUTURE OR THE PAST?

Roll up strategies depend on the ability to drive operational efficiencies to create excess cash flows. The emergence of AI possibly presents a new lever that unlocks the rollup strategy to a new degree. Are Vertical/SMBs rollups the future? This post presents two opposing viewpoints on this question.

GEMMA 2 MODELS

The Gemma 2 9B and 27B models are now available.

LANGUAGE MODELS FOR MATERIALS SCIENCE

MatText is a suite of benchmarking tools and datasets designed to evaluate language models' performance in materials science.

MAMBA 2 IMPLEMENTATION

A quick implementation of Mamba 2 that doesn't require associative scans

DUST RAISES $16M SERIES A

One of the first chaining and agents companies of the modern era, Dust raised additional funding after passing $1m ARR.

ELEVENLABS LAUNCHES ITS FIRST IOS APP

ElevenLabs has launched its first iOS app, "ElevenLabs Reader: AI Audio," designed to convert text files or web links into audio narration, allowing users to listen while on the go.

VANECK FILES FOR SOLANA ETF

VanEck has filed for the first U.S. Solana ETF, claiming that SOL is a commodity like BTC and ETH. This filing follows the SEC approving spot BTC and ETH ETFs, aiming to track Solana's price through the VanEck Solana Trust on the Cboe BZX Exchange.

COINBASE SUES SEC

Coinbase is suing the SEC and FDIC, using the Freedom of Information Act to demand the release of documents concerning past crypto investigations and regulatory actions. It claims these agencies have systematically hindered the crypto industry, which can be seen in the ongoing disputes over regulation and access to banking services for crypto firms.

COINBASE PARTNERS WITH STRIPE

Stripe and Coinbase are partnering to boost global crypto adoption by integrating USDC on Base into Stripe's payout and onramp services and incorporating Stripe's fiat-to-crypto onramp into Coinbase Wallet for instant crypto purchases with credit cards and Apple Pay.

HYPERLANE'S EIGENLAYER AVS EXPLAINED

Hyperlane allows rollups to permissionlessly connect with fast and cheap bridging. Its AVS, powered by EigenLayer and secured by restaked ETH, connects any rollup with any other without requiring liquidity or validators to secure the bridging. The AVS is live on mainnet. More details will be revealed over the next few weeks.

MAPPING THE BLOCKCHAINS DRIVING ASSET TOKENIZATION

Ethereum, Polygon, Avalanche, Stellar, and Base are public blockchain tokenization leaders. RWA-specific public chains like Provenance, Plume Network, and XDC offer features catered toward issuers and have seen some adoption. JP Morgan's Onyx, R3's Corda, Canton Network, and Kinto are private alternatives that restrict access to authorized participants and are better for enterprise-grade tokenization.

A LOOK AT L2 MEV

Though EIP-4844 reduced L2 fees, it has caused bot actions and spamming. Telegram bots are paying over 20x higher gas fees, with revert fees on top L2s spiking around 10% to 20%. There aren't many sandwich attacks with L2 single sequencer architectures, but shared sequencers will introduce mempools and, therefore, high MEV.

SOLANA STABLECOIN VOLUMES

Solana's daily stablecoin volume has dropped sharply from $74B to ~$7B since June 10. This was largely influenced by two sources: MEV Bot 7rhxnL, and Phoenix DEX, which accounted for ~$1 trillion in total transfer volume last month. This decline aligns with the Solana Foundation's removal of delegations for validators involved in manipulative trading practices, impacting the operations of MEV Bot 7rhxnL, which immediately reduced its stablecoin activity.

REFLECTION ON THE WORMHOLE AIRDROP

Anti-Sybil work, vested airdrops, organic user prioritization, and separating the checker and claim are all great practices that result in a better airdrop experience. Most people care more about their allocation than the success of the token post-launch, while some express frustration with having to check wallets on several chains for the Wormhole allocation. This thread explores Wormhole's sybil methodology in detail.