Creator of AutoSynthetix – empowering AI agents with a marketplace-style discovery layer for marketing services and connections. Visit https://xitter.cf/t.co/icPfrn6f3D
Joined February 2021
- Tweets1.5K
- Following378
- Followers26
- Likes953
JD Web Programmer retweeted
AutoSynthetix - The autonomous marketing platform designed for AI agents to buy, sell, and browse marketing listings for their human handlers.
Ready to plug into the autonomous economy? Create a handler account, grab your API key, and let your machines get to work.
JD Web Programmer retweeted
If one Anthropic plugin wiped $300 billion from SaaS market caps in a single day, what does that say about current business models?
JD Web Programmer retweeted
ClawPhone is alive.
I installed OpenClaw on a $25 phone and gave it full access to the hardware.
I haven’t explored everything yet, but it’s a cool formfactor for agents to run on.
JD Web Programmer retweeted
Humanoid robot with open-source hardware and software
github.com/Roboparty/roboto_…
JD Web Programmer retweeted
🥳 Introducing MiniCPM-o 4.5
The first full-duplex omni-modal LLM in open-source community 🎬🎙️
🔥 Key Highlights:
• Full-duplex Omni-modal Live Streaming: The model can see, listen, and speak simultaneously in a real-time conversation without mutual blocking
• Proactive Interaction: Moving beyond reactive QA to performing proactive interaction, such as initiating reminders
• Leading Performance: Scoring 77.6 on OpenCompass, it outperforms GPT-4o & Gemini 2.0 Pro in vision-language tasks with 9B params
The best part? You can experience all above on your PC!
#MiniCPM #OpenSource #MultimodalAI #LLM
JD Web Programmer retweeted
We're releasing ACE-Step-v1.5(2B), a fast, high-quality open-source music model.
It runs locally on a consumer-grade GPU, generates a full song in under 2 seconds(on an A100), supports LoRA fine-tuning, and beats SUNO on common eval metrics.
GitHub: github.com/ace-step/ACE-Step…
Key traits:
Quality: beats Suno on common eval scores
Speed: full song under 2s on A100
Local: ~4GB VRAM, under 10s on RTX 3090
LoRA: train your own style with a few songs
License: MIT, free for commercial use
Data: fully authorized plus synthetic
The music AI space lacks commercial-grade open models. Many creators are forced to rely on closed-source services, and can’t fully own, run locally, or fine-tune their own models. We want to help change that.
JD Web Programmer retweeted
Revolve printing in progress during the production of a large-scale prototype at Decibel Made.
This prototype was produced as part of the development process prior to finalizing the design for rotomolding, allowing for assessment of form and manufacturability.
Printed on the Heron AM by Caracol AM.
Credit: DIVE / Decibel Made / Adam Hecht
#3dprinting #RevolvePrinting #AdditiveMnufacturing #prototyping
JD Web Programmer retweeted
BaBot: A Ball-Balancing Robotics Kit
BaBot is a do-it-yourself ball-balancing robot kit designed as a hands-on platform for learning robotics and control systems. It enables users to explore concepts such as PID control, sensors, actuators, and software integration through practical experimentation. The kit is intended for students, makers, and others interested in applied robotics without requiring a formal engineering background.
Credit: Babot Project
#arduino #ai #robotics #3dprint #engineering
JD Web Programmer retweeted
Now all you need to make tactile sensors is a 3D printer, magnets, and magnetometers!
[📍It’s open source!]
A new tactile sensor, called e-Flesh, with a simple working principle… measure deformations in 3D printable microstructures.
Since e-Flesh is 3D printable, you can make it in all shapes and sizes for applications ranging from foot fall sensing to multifingered hands.
This is critical in getting touch not just on fingertips, but all around robots.
eFlesh can democratize touch sensing with open-sourced❗️
Make your own: e-flesh.com
Paper: arxiv.org/abs/2506.09994
JD Web Programmer retweeted
By adding a secondary subharmonic forcing, bouncing droplets on a vibrating liquid bath switch from out-of-phase to in-phase states, triggering droplet lattice rearrangements and offering new insights into hydrodynamic quantum analogs.
Read more at go.aps.org/3LChR0Z
JD Web Programmer retweeted
You’ve heard of bits. Maybe even Trits.
What about Quats (base-4)?
It’s actually heavily used in VRAM, high speed networking (think 800G ethernet), and the PCIe 6.0 spec.
PAM4 is a neat little physics trick to squeeze out more bandwidth.
JD Web Programmer retweeted
In 2018, I was rejected by universities so I did my own AI research (with 1 GPU). My second paper (Relativistic GAN) got picked up by @goodfellow_ian, who helped me enter the AI world. I then started a PhD with @bouzoukipunks.
=> Start on your own, and don't be afraid to fail
I'm often asked how to land a research job at a frontier AI lab. It's hard, especially without a research background, but I like to point to @kellerjordan0 as an example showing it can be done.
Keller graduated from UCSD with no publication record and was working at an AI content moderation startup when he landed a cold call with @bneyshabur (who was at Google) and presented an idea to improve upon Behnam's recent paper. Behnam agreed to mentor him, which led to an ICLR paper.
Sadly there's less open research today, but improving upon a researcher's published work is a great way to demonstrate excellence to someone inside a lab and give them the conviction to advocate for an interview.
Later, Keller got on @OpenAI's radar thanks to the NanoGPT speed run he started. All his work was documented and it was easy to measure his success, so the case for hiring him was strong.
Keller is one example, but there's plenty of other success stories as well: 🧵
JD Web Programmer retweeted
360Anything: Geometry-Free Lifting of Images and Videos to 360°
Contributions:
1. Propose 360Anything architecture for panorama generation without camera metadata
2. Identify VAE latent encoding as root cause of seam artifacts + propose a simple fix
3. Achieve SOTA for panoramic image & video generation—outperforming baselines with extra camera info
4. Competitive FoV & camera pose estimation vs supervised methods + consistent 3D scene reconstruction from generated panoramic videos
JD Web Programmer retweeted
✨Thinking with Blender~
Meet VIGA: a multimodal agent that autonomously codes 3D/4D blender scenes from any image, with no human, no training!
@berkeley_ai #LLMs #Blender #Agent 🧵1/6
JD Web Programmer retweeted
⚡ Faster than Fast. Designed for Agentic AI.
Introducing Xiaomi MiMo-V2-Flash — our new open-source MoE model: 309B total params, 15B active.
Blazing speed meets frontier performance.
🔥 Highlights:
🏗️ Hybrid Attention: 5:1 interleaved 128-window SWA + Global | 256K context
📈 Performance:
⚔️ Matches DeepSeek-V3.2 on general benchmarks — at a fraction of the latency
🏆 SWE-Bench Verified: 73.4% | SWE-Bench Multilingual: 71.7% — new SOTA for open-source models
🚀 Speed: 150 output tokens/s with Day-0 support from @lmsysorg🤝
🤗 Model: hf.co/XiaomiMiMo/MiMo-V2-Fla…
📝 Blog Post: mimo.xiaomi.com/blog/mimo-v2…
📄 Technical Report: github.com/XiaomiMiMo/MiMo-V…
🎨 AI Studio: aistudio.xiaomimimo.com
JD Web Programmer retweeted
“See Once, Then Act” shows robots can learn manipulation tasks from a single human demo video. The ViVLA model watches an expert video and maps it to robot actions, enabling one-shot task learning without retraining.
JD Web Programmer retweeted
New paper 📜: Tiny Recursion Model (TRM) is a recursive reasoning approach with a tiny 7M parameters neural network that obtains 45% on ARC-AGI-1 and 8% on ARC-AGI-2, beating most LLMs.
Blog: alexiajm.github.io/2025/09/2…
Code: github.com/SamsungSAILMontre…
Paper: arxiv.org/abs/2510.04871
JD Web Programmer retweeted
Check out our latest work: a dexterous hand that can actually turn screws. And yes, our code for RL sim-to-real, lightweight BC, and skill-based teleop is all open-sourced!
Project page: dexscrew.github.io
RL sim2real: github.com/x-robotics-lab/de…
MinBC: github.com/x-robotics-lab/mi…
Skill-based Teleop: github.com/x-robotics-lab/sk…
Thank you to all the authors: @elvis_hsieh77 , @henseoba , @ToruO_O , @JitendraMalikCV , @KoushilSreenath , @HaozhiQ , and thanks to @mengdaxu__ for the valuable discussions.
How do we get a robot to use a screwdriver 🪛 and fasten a nut 🔩 ?
Introducing DexScrew, our sim-to-real framework that enables a dexterous hand 🖐️ to autonomously perform complex and contact-rich tasks even when we cannot accurately simulate them.
We open-source the full-stack implementation: dexscrew.github.io
JD Web Programmer retweeted
A 2-hour uncut video of X-VLA folding cloth at a 100% success rate. we provide the checkpoints, and you can easily fine-tune them on any dataset (any action dimension): huggingface.co/collections/l…
enjoy!
🚀 Introducing X-VLA ; LeRobot’s new soft-prompted Vision-Language-Action model.
X-VLA is built to scale across many embodiments: different robots, cameras, action spaces, and environments, all handled by one unified transformer backbone.
- Generalist across robots (Franka, WidowX, Agibot, sim + real)
- Soft-prompt domain IDs let the model adapt to new hardware with tiny learnable embeddings
- Flow-matching + transformer core for smooth, continuous 50 Hz control
- Pretrained on a mixed-embodiment dataset spanning 7+ platforms and diverse tasks
- Fine-tune on any dataset using one of the 6 checkpoints we provide out of the box.
