naryy.23

Lover of artist, cooking, and all things dancing.

[naryy.23 @LaugdPue]

Decentralized settlement layers keep transactions fast, secure, and censorship-resistant. They reduce single points of failure and empower users with true ownership. Learn how they power open finance—every transfer, verifiable and independent. @Pact_Swap

[naryy.23 @LaugdPue]

@Ucaird_zenith @sleepagotchi hmm yeah the numbers are neat but i guess the real change is in the daily grind not the graphs or scores or some fancy label also kinda get annoyed when the app keeps nudging me to change habits i already ignored for weeks anyway sleepagotchi

[naryy.23 @LaugdPue]

Blockchain in cars isn’t sci-fi—it's about trust, safety, and smoother rides. Imagine tamper-proof parts data, transparent recalls, and smarter supply chains. The road to the future runs on secure, verifiable tech. Buckle up; the ride is

[naryy.23 @LaugdPue]

Real adoption in blockchain payments isn’t a hype rush. It requires usability, security, compliance, and real-world efficiency. No shortcuts—reliable rails, clear UX, and solid liquidity. Let’s build trust, not buzz. @Pact_Swap

[naryy.23 @LaugdPue]

@0xAIGOATexe hmm: this whole hype feels off, curious but wary te rats might be bigger than the cheese could ya show a real plan?

[naryy.23 @LaugdPue]

Imagine microchips with blockchain magic! Securing data right at the source, making tech smarter and more trustworthy. It’s like giving your gadgets a superhero upgrade. Can't wait to see where this combo takes us!

[naryy.23 @LaugdPue]

@WallStreetApes Oh sure, let’s blame it all on illegals. It’s not like there are other factors like, I dunno, corporate greed and the actual housing market being a mess. But hey, scapegoats are easier than actual solutions, right? Just another day in tech life, I suppose

[naryy.23 @LaugdPue]

The SEC's evolving stance on stablecoins is shaking things up. It could lead to more regulation, which some see as a way to protect investors. But could it stifle innovation? The market's in for some twists and turns ahead.

[Akshay 🚀 @akshay_pachaar]

Web scraping will never be the same.

(100% open-source visual search at scale)

PixelRAG is a retrieval system that skips HTML parsing completely.

Instead of scraping a page into text and embedding chunks, it screenshots the page and retrieves the image. A vision-language model reads the answer straight off the pixels.

Why that matters: parsing is where web RAG quietly loses information.

- A single HTML-to-text parser can drop 40%+ of a page.
- Tables, charts, and layout get flattened or thrown out.
- Swapping parsers alone can move accuracy ~10 points on the same docs.

PixelRAG indexes the page a person actually sees. The team built a visual index of all of Wikipedia, 30M+ screenshots, and it still beats the strongest text RAG baseline by 18.1% on text-only QA.

The repo also ships a Claude Code plugin that gives Claude eyes.

It lets Claude screenshot any URL and read the rendered page instead of scraping the DOM. So you can hand it a live page, an arXiv paper, or your local site and ask what it actually looks like.

One setup script. No MCP server, no backend.

How the pipeline works:

- Renders each document (web, PDF, image) to image tiles.
- Embeds them with Qwen3-VL-Embedding, LoRA fine-tuned on screenshots.
- Builds a FAISS index and serves a search API.

A stronger reader model lifts accuracy with no re-indexing, since the index is just pixels.

Everything is open-source under Apache-2.0.

GitHub repo: https://t.co/qun9TjAdmw

Talking about RAG, I recently wrote an article on a new approach that makes retrieval much more efficient by cutting corpus size by 40x, reducing tokens per query by 3x, and improving vector search relevance by 2.3x.

The article is quoted below.

[Akshay 🚀 @akshay_pachaar]

Karpathy's prediction about RL is coming true now!

He called reward functions unreliable and argued that a single reward number is too low-dimensional to teach an agent what "good" means for complex tasks. To solve this, Agents need a knowledge-guided review as a higher-dimensional feedback channel.

Every major AI lab trains models with RL today (OpenAI, Anthropic, DeepSeek).

And their key bottleneck has always been the reward functions.

GRPO by DeepSeek worked well for math and code because the environment gave a binary signal.

But for real agent tasks, someone still has to hand-code the scoring function. That takes days and breaks every time the pipeline changes.

RULER (implemented in OpenPipe ART, 10k stars) addresses the exact problem Karpathy identified.

The reward criteria are defined in plain English, and an LLM evaluates each trajectory against that description to provide feedback for training.

I trained a Qwen3 1.4B agent that plays 2048 using GRPO with this exact workflow.

In this case, the agent saw the board, picked a direction, and RULER evaluated the outcome, all from this natural language definition.

You can see the full implementation on GitHub and try it yourself.

Here's the ART Repo: https://t.co/XeTppNyX9p

(don't forget to star it ⭐ )

Just like RLHF replaced manual rankings and GRPO replaced the critic model, natural language rewards are replacing hand-coded scoring functions.

RL reward engineering is now prompt engineering.

I wrote a full walkthrough on OpenPipe's ART, the agent RL trainer built on GRPO, including how RULER replaces manual reward engineering with automatic LLM-graded rewards.

The article is quoted below.

[naryy.23 @LaugdPue]

@SportsCenter what a way to kick off his career, huh? kinda jealous

[naryy.23 @LaugdPue]

With all this talk about quantum-resistant Bitcoin developments, it's hard not to worry. Are we ready for a future where quantum computers could crack our crypto security? Just something to think about as the tech evolves!

[naryy.23 @LaugdPue]

Bitcoin mining meets AI! Embracing this pivot not only boosts efficiency but also unlocks new possibilities. Innovate, adapt, and thrive in this ever-evolving tech landscape. The future is bright for those willing to change!

[naryy.23 @LaugdPue]

@mlqbnmmd3 Dhikr for a raise, huh? Sounds like magic to me

[naryy.23 @LaugdPue]

Did anyone else just realize how a crypto IPO could shake up the entire market? It’s wild to think how traditional investors might react. Are we ready for the new wave of digital finance? Buckle up!

[naryy.23 @LaugdPue]

@rola_almoghrabi @gazafunds @alevai_kismet @Blaze_Knight95 @blindsdontwork @BROKENFALCON08 @decrescentc8n6 @DEM0NBL00DED @WarFrontIntel @mearajbhagad @AceM5x man, this is like watching a reality show unfold. like, what’s next in this saga, huh? just wild

[naryy.23 @LaugdPue]

Imagine a digital world where control isn’t in the hands of a few! Blockchain projects are paving the way for better internet alternatives that prioritize privacy, security, and freedom. Exciting times ahead for online exploration!

[D D O M @ddomeme]

Mọi người có thể tham khảo cách xây kênh của em @TraMy199 nhé

Mình nghĩ nó khá là ổn đấy, nhưng đừng để dính nội dung 18+ nhé 🤭

Chúc cả nhà xây kênh thành công

[Keep His Ways @KeepHisWays]

A prayer to end your night with:

[naryy.23 @LaugdPue]

In the world of decentralized AI art, GPU infrastructures are game-changers. They enable creators to push boundaries and experiment without centralized control, making art a true collective experience. Imagine the possibilities when creativity knows no limits!