Gabriel

@gostack.bsky.social

391 Followers 295 Following

• Ending Bad Ideas Forever® • 100 y/o in internet yrs • 🇵🇷 Raised in The Bronx • Start-up Techie • Ex-Pat Scot 🏴󠁧󠁢󠁳󠁣󠁴󠁿 • Ex-Pat Dutchie 🇳🇱 • Prefers bikes • Polyglot - EN/SP/NL & Celtic Languages • Worker First • 🚫 Crypto

925 Irish words learned

Viral post mocks anti-AI critics.
Replies overwhelmingly say they never used AI tools in the first place.

People use "frankly," "honestly," and "truthfully" as conversational tools (known as sentence modifiers or discourse markers) to signal the vibe of what they are about to say. While they have overlapping definitions, they are not exact duplicates.The Nuanced Differences Between the WordsFrankly: This word carries an edge of bluntness, directness, or a lack of sugar-coating. When someone says "Frankly," they are usually delivering an unpleasant truth, a harsh critique, or an opinion they expect you might not like.Example: "Frankly, I think this movie is terrible."Honestly: This word focuses on sincerity and emotional authenticity. It is a neutral modifier used for both positive and negative statements, or simply to add emphasis.Example: "Honestly, I really appreciate your help." or "Honestly, I don't care either way."Truthfully: This word is rooted heavily in factual accuracy. It is often used to signal that the speaker is answering a question accurately without omitting details, sometimes defensively if they think they won't be believed.Example: "Truthfully, I don't know where the keys are."

I completed the dutch course on Duolingo.

Headline that reads:
Meet the Zillennials: The luckiest micro-generation in the workforce, born between 1993 and 1998.

A terracota woman sculpture on a crucifix.

A picture of an ad for a robot puppy.

Lovable's Servers: Lovable hosts its own platform and infrastructure entirely on AWS.The Coding AI (Claude): When you type a prompt into Lovable to build a feature, Lovable routes that prompt to Anthropic's Claude models, which act as the main "brain" generating your app's code.The Gemini Integration: Lovable uses Gemini as an optional, user-facing feature. It provides a built-in "AI Connector" so that the apps you build can easily use Gemini for tasks like handling customer service chats or pulling data from Google Workspace.In short: Lovable is hosted on AWS, powered by Claude to write code, and offers Gemini as a tool for the apps it creates.

The boat we took to Inchcailloch

The summit of Inchcailloch in Scotland.

A drawing of an elephant that was swallowed by a snake both inside and out.

My homemade electric bike with two milk crates stacked in the back.

Countries with Unrestricted Birthright Citizenship
The following countries grant citizenship to nearly anyone born on their soil, with common exceptions only for children of foreign diplomats:

North America: United States, Canada, Mexico.
Central America: Belize, Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, Panama.
South America: Argentina, Bolivia, Brazil, Chile, Ecuador, Guyana, Paraguay, Peru, Uruguay, Venezuela.
Caribbean: Antigua and Barbuda, Barbados, Cuba, Dominica, Grenada, Jamaica, Saint Kitts and Nevis, Saint Lucia, Saint Vincent and the Grenadines, Trinidad and Tobago.
Africa: Lesotho, Tanzania, Chad (unrestricted, but child must opt-in at age 18).
Oceania: Fiji, Tuvalu.
Asia: Pakistan.

Countries with Restricted Birthright Citizenship
Many other nations offer a restricted form of jus soli, where citizenship is granted only if certain conditions are met, such as one parent being a legal resident or the child living in the country for a set number of years. Examples include:

United Kingdom: At least one parent must be a citizen or "legally settled".

Australia: Granted if one parent is a citizen/permanent resident, or if the child lives there for 10 years.

Germany: At least one parent must have had a permanent residence permit and lived in Germany for several years (currently 5-8 years depending on recent updates).

France: Often granted upon reaching adulthood if the child has lived in France for at least five years since age 11.

Pizza with anchovies, red onions and green peppers.

Bernie Sanders: The Vermont Senator was a long-time ally, endorsing Fetterman during his 2018 run for Lieutenant Governor and again during his 2022 Senate campaign.

Bob Casey: Pennsylvania’s senior Senator strongly vouched for Fetterman, particularly defending him following his stroke during the 2022 campaign.

Alexandria Ocasio-Cortez: The U.S. Representative from New York endorsed Fetterman during his 2022 Senate race, though Fetterman noted at the time that he would not formally join "The Squad".

Martin O'Malley: The former Maryland Governor and presidential candidate endorsed Fetterman during his initial 2016 Senate bid.

Oprah Winfrey: While not a career politician, the influential media icon and Democratic supporter endorsed Fetterman over his opponent, Dr. Mehmet Oz, in the final days of the 2022 election.

A picture of Dr. Emmet Brown, inventor of time travel.

Gemma3's response:
According to the provided text, Bronxnet, along with other New York community media centers, is facing a funding crisis due to declining cable subscriptions and the rise of streaming services. Cable subscriptions have decreased significantly, impacting the revenue Bronxnet receives.

To address this, Bronxnet is advocating for the Community Media Reinvestment Act. This legislation would create a fund generated from taxing streaming services (like Netflix, Hulu, and Amazon) to reinvest in community media centers. If passed, this fund could generate between $250 million to $500 million for New York media organizations like Bronxnet.

The legislation aims to provide a lifeline for Bronxnet and other community media centers, enabling them to continue providing media education, production tools, and airtime to community members.

ChatGPT explaining how you can start making a flux capacitor in real life.

The phrase "up jump the boogie" wasn't just a random line—it was a vocal technique designed to mimic the energy of a live DJ performance.
1. The "Vocal Drum Roll"
Wonder Mike, the member of The Sugarhill Gang who performed the opening verse, described the entire sequence ("To the bang-bang boogie, say up jump the boogie...") as a vocal drum roll.
The Goal: He wanted to create a percussive sound using the letter "B" to match the driving rhythm of the beat.
The Sound: It was meant to function like a "break" in the music, where the voice provides the same explosive impact as a drummer hitting a snare or a DJ dropping a fresh record.
2. Deep Roots in "Boogie" Culture
While the specific phrasing became legendary through "Rapper's Delight," the components of the lyric have older roots:
The "Boogie": The term "boogie" has a century-long history, starting with the boogie-woogie piano style of the 1920s and 1930s. By the 1970s, it was the universal slang for both disco music and dancing.
Live MC Origins: Phrases like "up jump the boogie" and "you don't stop" were common crowd-participation chants used by early Bronx MCs like Lovebug Starski and Keef Cowboy. They were used to "hype" the crowd and keep the energy high during long DJ sets.

Workers, Managers,
and Welfare Capitalism
The Shoeworkers and Tanners
of Endicott Johnson,
1890-1950
GERALD ZAHAVI
University of Illinois Press

my A.I. says that I'll increase the speed of my LLMs thusly:
Speed and Generation
TensorRT-LLM is built for maximum speed. While Ollama is very fast (delivering about 60–80 tokens per second), TensorRT-LLM can nearly double that to 140–180 tokens per second on your hardware. It achieves this by "compiling" the model into a rigid, highly efficient binary that uses your GPU’s Tensor Cores more effectively than the general-purpose engine used by Ollama.

Memory and Intelligence
Ollama uses a standard approach to memory that is simple but can be wasteful. TensorRT-LLM uses a technology called PagedAttention, which manages your 16GB of VRAM like a computer's RAM. It breaks data into small "pages," filling every available gap of memory. This allows you to handle much longer conversations or "chunk" massive documents without the model slowing down or running out of memory.

Setup and Flexibility
The biggest difference is in the "effort" required. Ollama is "Plug and Play"—you download a model and it works instantly. TensorRT-LLM requires an upfront "build" process that optimizes the model for your specific 4070 Ti Super. This build takes 10–20 minutes, but you only do it once. Additionally, while Ollama runs almost any model found online, TensorRT-LLM is more selective, supporting mainly the popular "heavy hitters" like Llama 3, Mistral, and DeepSeek.

Summary for Your Setup
On an RTX 4070 Ti Super, the switch to TensorRT-LLM is effectively a "Pro" upgrade. You trade the convenience of instant model swapping for a significant boost in professional-grade performance and memory efficiency.

blueprint for my new tiny house.