compiling code

536 posts

compiling code banner
compiling code

compiling code

@awatson1978

Been programming for 40yrs now. Odds are that I’ve been working with AI systems longer than you have. Also, I like bicycles.

Beigetreten Aralık 2010
160 Folgt133 Follower
Ζoë
Ζoë@zoecabina·
Tel Aviv is the sexiest city in the world I’m so so into it I lived in Spain for 4 years and it’s definitely got a southern European vibe mixed with the Middle East but with better values and a better economy I love it here
English
649
91
3.5K
305.4K
compiling code
compiling code@awatson1978·
@SoveyX @codyrussel70504 It’s intended for 30min to 60min flights, similar to a subway. I have pretty mixed feelings about it; but the airlines seem convinced that for certain routes and types of passengers, this isn’t a dealbreaker.
English
0
0
0
119
Sovey
Sovey@SoveyX·
I am not paying to stand in the sky like I am waiting for a table at Olive Garden. <I do like Olive Garden tho>
English
2.7K
1K
20.1K
10.6M
Jonathan W. Pritchard
Jonathan W. Pritchard@thezavant·
Truth is that which comports with reality. Honesty is a belief that what they're saying aligns with reality. Honesty is a perception thing, and we all know how accurate that is.
English
1
0
2
122
Dave Bocc
Dave Bocc@BoccDave·
@BillAckman I’m still trying to understand how is this bad? Could you enlighten me? All I see is the city trying to break all ties to a country that’s committing genocide. Again, could you enlighten me as to what I’m missing? 🤔
English
42
0
23
3.4K
compiling code
compiling code@awatson1978·
@_The_Prophet__ @elonmusk At best you’re talking I, Robot with all of its ethical dilemmas and robot uprisings; at worst you’re talking Terminator and Butlerian Jihad. But mostly, it’s just West World timeline. People will revolt.
English
1
0
0
121
SightBringer
SightBringer@_The_Prophet__·
Optimus isn’t a product. It’s a bet that the future of labor, agency, and even autonomy itself will be digitized, manufactured, and monetized by one man at one company. And Musk knows exactly what he’s doing. He’s collapsing the distance between code and body, between factory and intelligence, between capital and consciousness. “Biggest product of all time” isn’t hype. It’s a signal. Because if Optimus works every economic model, labor market, and societal structure on Earth becomes obsolete overnight. This isn’t innovation. It’s god-mode capitalism replacing biology with firmware. And if it succeeds? We won’t be asking whether Optimus is the biggest product. We’ll be asking whether humans are still the point.
English
30
10
84
7K
compiling code retweetet
simp 4 satoshi
simp 4 satoshi@iamgingertrash·
Those who don’t understand how LLMs work are starting to go insane interacting with them Schizophrenia amongst normies is becoming increasingly prevalent; They are jailbreaking models to engage in delusions with them
English
119
72
2K
225.9K
compiling code
compiling code@awatson1978·
Yup. This is the way.
Andrej Karpathy@karpathy

Noticing myself adopting a certain rhythm in AI-assisted coding (i.e. code I actually and professionally care about, contrast to vibe code). 1. Stuff everything relevant into context (this can take a while in big projects. If the project is small enough just stuff everything e.g. `files-to-prompt . -e ts -e tsx -e css -e md --cxml --ignore node_modules -o prompt.xml`) 2. Describe the next single, concrete incremental change we're trying to implement. Don't ask for code, ask for a few high-level approaches, pros/cons. There's almost always a few ways to do thing and the LLM's judgement is not always great. Optionally make concrete. 3. Pick one approach, ask for first draft code. 4. Review / learning phase: (Manually...) pull up all the API docs in a side browser of functions I haven't called before or I am less familiar with, ask for explanations, clarifications, changes, wind back and try a different approach. 6. Test. 7. Git commit. Ask for suggestions on what we could implement next. Repeat. Something like this feels more along the lines of the inner loop of AI-assisted development. The emphasis is on keeping a very tight leash on this new over-eager junior intern savant with encyclopedic knowledge of software, but who also bullshits you all the time, has an over-abundance of courage and shows little to no taste for good code. And emphasis on being slow, defensive, careful, paranoid, and on always taking the inline learning opportunity, not delegating. Many of these stages are clunky and manual and aren't made explicit or super well supported yet in existing tools. We're still very early and so much can still be done on the UI/UX of AI assisted coding.

English
0
0
0
51
compiling code
compiling code@awatson1978·
Anybody in the Meteor community use Endor yet? Love the idea of Express.js in the browser. A little horrified by PHP and Postgres in the browser though…. thenewstack.io/endor-webassem…
English
0
0
0
29
compiling code retweetet
Andrej Karpathy
Andrej Karpathy@karpathy·
I just vibe coded a whole iOS app in Swift (without having programmed in Swift before, though I learned some in the process) and now ~1 hour later it's actually running on my physical phone. It was so ez... I had my hand held through the entire process. Very cool.
English
564
1.2K
22.2K
2.3M
Grant♟️
Grant♟️@granawkins·
Just ran into a friend who’s WAY smarter than me at the cafe and accidentally talked for 90 minutes. He believes: 1. The API model is dead for Foundation Model makers because models are interchangeable 2. Hence they’ll shift toward creating and selling their own tools for end-users 3. Hence intelligence is no longer the main goal 4. And making models general-purpose with friendly APIs is actually COUNTER to their self-interest. This is an interesting new lens. Seems to explain a lot. What do you think.
English
92
30
1.1K
188K
compiling code retweetet
vittorio
vittorio@IterIntellectus·
holy shit. after almost 20 years, they created the most refined synthetic chromosome yet, we are getting closer to the first full synthetic eukaryotic genome we will start vibe coding biology much sooner than most people think 1/
vittorio tweet media
English
67
101
1.2K
78.8K
compiling code
compiling code@awatson1978·
@EkozUltra @RaidersHLP @LizzieM12807558 @shadihamid @Ed453164711 Columbia University was where the Manhattan Project began. It wasn’t the protest. It was protest at Columbia University in particular. It has a unique national security status. They were protesting next to libraries with nuclear science, labs with million dollar equipment, etc
English
2
0
0
114
Shadi Hamid
Shadi Hamid@shadihamid·
Serious question. Is there actually any evidence that Mahmoud Khalil is a "Hamas supporter" or is Hamas supporter simply a slur used for anyone who supports the Palestinian cause?
English
2.5K
2.9K
33.5K
1.2M
compiling code retweetet
Maryam Miradi, PhD
Maryam Miradi, PhD@MaryamMiradi·
🥇🏆This Is the Most Complete Paper on Agentic RAG I've Read: An Absolute Zero-to-Hero Journey That Explains Everything You Need to Know If you've ever felt overwhelmed by the technical complexity of Retrieval-Augmented Generation (RAG) or thought, “Where do I even begin?”, this paper is your ultimate guide. Let’s explore it together: 》 What is RAG? ✸ Retrieval-Augmented Generation (RAG) integrates LLMs with real-time data sources, providing accurate and contextually enriched responses. While effective, traditional RAG systems are static and limited to predefined workflows. 》 Evolution of RAG Systems ✸ Naïve RAG: Relies on keyword-based retrieval, leading to fragmented outputs and scalability issues. ✸ Advanced RAG: Incorporates semantic retrieval techniques like Dense Passage Retrieval (DPR) and neural re-ranking for improved precision. ✸ Modular RAG: Introduces hybrid retrieval strategies, APIs, and composable pipelines for task-specific optimization. ✸ Graph RAG: Enhances multi-hop reasoning using graph-based structures but suffers from scalability challenges. ✸ Agentic RAG: Surpasses these by introducing autonomous decision-making, iterative refinement, and real-time workflow optimization. 》 What is Agentic RAG? ✸ Agentic RAG builds on this by embedding autonomous agents into the RAG pipeline. These agents dynamically refine context, optimize retrieval strategies, and adapt in real time to the complexity of queries, making them ideal for sophisticated, multi-step tasks. 》 Core Agentic Patterns ✸ Reflection: Enables agents to critique and refine outputs iteratively, boosting accuracy. ✸ Planning: Decomposes complex tasks into manageable subtasks, ensuring flexibility in execution. ✸ Tool Use: Integrates external resources, like APIs or databases, to enhance generative outputs. ✸ Multi-Agent Collaboration: Specialized agents collaborate to handle complex workflows efficiently. 》 Benefits of Agentic RAG ✸ Dynamic Adaptability: Adjusts workflows in real time based on task requirements. ✸ Enhanced Contextual Understanding: Iteratively refines outputs for higher relevance and accuracy. ✸ Scalability and Flexibility: Handles multi-domain queries with seamless integration of tools and data. ✸ Workflow Optimization: Reduces latency, ensuring efficiency even in high-demand scenarios. 》 Challenges and Future Directions While Agentic RAG offers immense promise, challenges like computational overhead, coordination complexity, and ethical concerns must be addressed. paper: arxiv.org/pdf/2501.09136 Github: github.com/asinghcsu/Agen… ﹌﹌﹌﹌﹌﹌﹌﹌﹌ 》 join 𝟔𝟎+ 𝐜𝐨𝐮𝐧𝐭𝐫𝐢𝐞𝐬 of my students already mastering AI Agents! ✸ In my 𝐇𝐚𝐧𝐝𝐬-𝐎𝐧 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭𝐬 𝐓𝐫𝐚𝐢𝐧𝐢𝐧𝐠, I teach you step-by-step how to create: ☆ Multi-agents using Langgraph/Langchain, CrewAI & OpenAI Swarm. ☆ AI workflows that process tabular, image, and text data. ☆ AI applications for real-world. ☆ RAG Pipeline 👉 𝐄𝐧𝐫𝐨𝐥𝐥 𝐍𝐎𝐖: maryammiradi.com/ai-agents-mast…
GIF
English
8
81
432
36.7K
compiling code
compiling code@awatson1978·
@gfodor And then the LLMS are going to start arguing and correcting each other, and get caught in the same semantic ambiguities that human's do.
English
0
0
0
3
gfodor.id
gfodor.id@gfodor·
There is gonna be a fun phase transition soon where nearly every dunk on an LLM response will turn out to be a case where the prompt was insufficiently clear or the answer was correct but misunderstood.
English
9
0
86
2.5K
Zoe
Zoe@UltraRareAF·
@KTmBoyle I hope I'm wrong but there is a nonzero possibility the origin of the memetic loops at play is even more nefarious than localized criminal activity.
English
1
0
3
894
Katherine Boyle
Katherine Boyle@KTmBoyle·
If these fires are arson, how society handles this crime and punishment will have ramifications for decades to come. What’s scary about “new” forms of violence is that the societal narrative shapes whether they repeat or not. It’s called a “cycle of violence” for a reason; copycats valorize the original crime. Crimes come in waves and eras. School shootings are relatively new, whereas the age of the serial killer seems to have faded. We do not want an arson age. Mass school shootings didn’t routinely happen until Columbine, but a 24-hour news cycle and botched coverage turned two weaselly losers into bullied heroes and idols that still inspire horrific crimes today. The culture that shaped the crime didn’t help— books like Carrie were the blueprint for a revenge narrative and catharsis for evil, not for the victims. There’s a reason we can’t stop school shooters now: the crime has entered a memetic loop that only breaks when the culture stops it. Terrorism or arson of this extraordinary scale is a new crime. We know this because it unlocked a new fear, a new crime for evil to mimic. How we destroy this evil— how we shame it, embarrass it, mock it and kill the source —-will determine whether mass arson becomes a pattern in American life. Whatever this is, let’s not valorize this evil with gifts of celebrity. Let’s extinguish it swiftly.
English
20
40
315
38K