Sabitlenmiş Tweet
Vin ./
1.2K posts
Vin ./
@VinoisAsian
ALL IN | MAXI | @Gradient_HQ
Taiwan Katılım Ekim 2024
291 Takip Edilen364 Takipçiler
We're expanding pre-train and model size, it's time to explore post training.
like how we, human learn, experience, we learn as we do , on the go, everyday!
Gradient@Gradient_HQ
Benchmarks that test what models have memorized are saturating fast. ARC-AGI-3 is asking a harder question: can AI actually learn something new on the fly? One direction we've been exploring: multi-agent orchestration. In our study, coordinating four frontier LLMs across multiple turns consistently matched or outperformed the strongest single model, even on tasks none of them could solve alone. The gap between "best single model" and "best coordination of models" is where a lot of the real progress is hiding. More on our multi-turn, multi-agent orchestration study: arxiv.org/abs/2509.23537
English
Qwen3.5 0.8B running real-time video captioning on a Mac Studio M2 Ultra.
<1s per frame.
269 frames from a 3m49s video.
Streaming descriptions as it plays.
Pause anywhere, it actually understands the scene.
~1GB model.
Local AI is getting unreasonably capable.
Video credit: @stevibe
English
Vin ./ retweetledi
Vin ./ retweetledi
Our GTC takeaway is clear: NVIDIA is betting hard on open.
- NemoClaw turns OpenClaw into enterprise infrastructure.
- Nemotron 4 will be open-sourced.
- Nemotron Coalition puts eight labs on a shared open frontier model.
This is what we've been building toward. Open infrastructure for open intelligence is the direction the biggest AI companies are taking.
English
everyone loves posting profits
but here’s the other side of crypto
few years ago i tried to build like a
"bullish portfolio with huge potential"
results are on the pic
honestly i don’t even want to calculate the total loss anymore haha
but that tuition fee ❗️
glad that phase is behind me
feel free to see your same story here
English
Vin ./ retweetledi
Every AI model you use went through two phases:
- Pre-training builds raw intelligence (reading the internet).
- Post-training builds judgment (learning to be useful).
Reinforcement learning plays a huge part in the latter.
Here's why it matters and how we make it better.
English
Vin ./ retweetledi
Vin ./ retweetledi
perplexity just announced always-on ai running on a mac mini.
the category is real. the question is whether your always-on ai should phone home to someone else's servers or run entirely on your own hardware.
we built parallax for the second option.
Perplexity@perplexity_ai
Announcing Personal Computer. Personal Computer is an always on, local merge with Perplexity Computer that works for you 24/7. It's personal, secure, and works across your files, apps, and sessions through a continuously running Mac mini.
English
Welcome to Clawbox
> No need to have a Macmini or renting VPS.
> No setup hassle. just download and run.
> Various model options at arguably the most affordable prices.
Get it here : commonstack.ai/clawdbot
English
Morning CT 🌤️
Lately it feels like a lot of projects have quietly disappeared
Teams shutting down.
Communities fading away.
Builders suddenly finding themselves without a place to keep pushing their ideas forward
And if you’ve been around this space long enough, you know that strange feeling
Because when a project ends, the curiosity doesn’t disappear.
The ideas don’t disappear.
And the people who genuinely love building definitely don’t disappear
They just start looking for a new place where that energy still exists
That’s something I’ve started to notice about @Gradient_HQ
It doesn’t really feel like a typical project community.
It feels more like a group of researchers, developers, and curious minds who are genuinely interested in exploring intelligence together. People who are still asking questions about where AI is going and what we can build next.
In a space that often moves from hype cycle to hype cycle, it’s refreshing to find communities where people still care about the process of experimenting, learning, and sharing ideas.
After a while, it starts to feel less like “joining a project”
and more like finding people who are on the same journey.
Almost like a small family of builders.
Because at the end of the day, communities like this aren’t just about conversations.
They’re about the things people are building together.
Experiments around decentralized AI systems, research on distributed reinforcement learning, and infrastructure like Echo-2, which is trying to rethink how reinforcement learning can scale across distributed compute.
But none of that really works without the people behind it.
And maybe that’s what makes @Gradient_HQ feel different.
It’s not just a project.
It’s a group of people who are still curious enough to keep building.
If you're still exploring AI, still experimenting, still trying to understand where intelligence might go next…
You’ll probably feel at home here
@VinoisAsian @HexxRL @realsirandrew @Leila1113514 @miketwinks
English
what crazier is the device receive a 200% performance boost at any crowded coffeshop
Josh Kale@JoshKale
Apple just dropped the M5 Max MacBook Pro and it's an AI Powerhouse. 4x faster AI Compute over M4 Max. These Specs are insane: - 18-core CPU with 6 "super cores" = world's fastest CPU core - 40-core GPU = rivals an RTX 4070 in a laptop - 128GB unified memory = more than most servers - 614 GB/s bandwidth = 4x what a DGX Spark gets - 24-hour battery life You can now run Llama 70B, a model that required a $40,000 GPU cluster 18 months ago on, a laptop at your local coffee shop. At ~20-30 tokens/sec it's fast enough to actually use. The "local AI" revolution just shipped for $3,499.
English
Vin ./ retweetledi
Great conversation with @supercyclepod.
Building Gradient has been a relentless process, but the real inflection point is ahead. Intelligence belongs to everyone, and we intend to make that a reality.
Watch it in full:
Supercycle@supercyclepod
AI should be a public good, not something gatekept by a handful of megacorps We had Eric Yang, co-founder of Gradient Network, on the pod this week to talk through exactly that. Gradient's "Open Intelligence Stack" includes: i) Parallax for distributed model serving ii) Echo for decentralized reinforcement learning The whole thesis is that anyone should be able to run large models on consumer hardware (yes, including your Mac Minis + OpenClaws) Eric breaks down their $10M seed round led by Pantera, Multicoin, and HSG; where he sees the industry heading; and why post-training is going to be the dominant force in enterprise. Timestamps: 00:00 Intro 01:15 AI market is booming 02:29 Local compute is a hot topic 03:02 Parallax Inference Engine 04:34 Intelligence as a public good 05:46 AI models will become a commodity 07:32 Bottlenecks in AI models accessibility 09:34 Smaller AI models are catching up 11:01 How Gradient's Infrastructure Enables Model Development 12:15 Model post-training 14:24 How does reinforcement learning work? 17:35 AI going rogue 19:20 Gradient's token 23:02 AI entrepreneurs that Eric admires 26:11 Use cases on chain for AI 31:34 The trade-offs of coming to crypto 35:09 How low-spec GPUs will work on Gradient Ecosystem 38:08 Post-training will be the dominating force for enterprise 38:43 Open source models are way cheaper 41:39 Eric's founding story 49:07 Empowering researchers globally 53:37 Why did Multicoin Capital and Pantera Capital invested in Gradient 55:08 One-click deploy agent 58:16 Gradient in 3 years
English
these valuations should not be trusted.
also such low volumes of 250k usd between the two yet its supporting nearly “1.55T USD” in valuation is just ludicrous.
anyone promoting these is trying to find someone gullible.
these do not give you ownership in OpenAI or Anthropic, and at these valuations you aren’t getting private asset returns (as it’s already long gone) for private asset risks.
English
Someone just built a 3D thought map by turning Obsidian embeddings into living networks.
It lets you see the "shape" of your mind.. centralized, decentralized, and distributed.
English
Vin ./ retweetledi
AI should be a public good, not something gatekept by a handful of megacorps
We had Eric Yang, co-founder of Gradient Network, on the pod this week to talk through exactly that.
Gradient's "Open Intelligence Stack" includes:
i) Parallax for distributed model serving
ii) Echo for decentralized reinforcement learning
The whole thesis is that anyone should be able to run large models on consumer hardware (yes, including your Mac Minis + OpenClaws)
Eric breaks down their $10M seed round led by Pantera, Multicoin, and HSG; where he sees the industry heading; and why post-training is going to be the dominant force in enterprise.
Timestamps:
00:00 Intro
01:15 AI market is booming
02:29 Local compute is a hot topic
03:02 Parallax Inference Engine
04:34 Intelligence as a public good
05:46 AI models will become a commodity
07:32 Bottlenecks in AI models accessibility
09:34 Smaller AI models are catching up
11:01 How Gradient's Infrastructure Enables Model Development
12:15 Model post-training
14:24 How does reinforcement learning work?
17:35 AI going rogue
19:20 Gradient's token
23:02 AI entrepreneurs that Eric admires
26:11 Use cases on chain for AI
31:34 The trade-offs of coming to crypto
35:09 How low-spec GPUs will work on Gradient Ecosystem
38:08 Post-training will be the dominating force for enterprise
38:43 Open source models are way cheaper
41:39 Eric's founding story
49:07 Empowering researchers globally
53:37 Why did Multicoin Capital and Pantera Capital invested in Gradient
55:08 One-click deploy agent
58:16 Gradient in 3 years
English
Vin ./ retweetledi
@justin_solstice they hate it cause it makes them look like years of trying is wasted.
A dude with $20 subscription can outperform them in matter of seconds.
What they need is to adapt and work with it they got better foundation so they should be the one who benefit from it more smh
English
A lot of vibecoder hate on X from devs. Open challenge to any devs. If you feel your code and skills are superior, I am happy to 1 v 1 on your choice of language, and application.
English