Pengfei Zhang

CS Ph.D. graduated in UMass Lowell. Computer Vision, Video Action Detection, Medical Image, Deep Learning, Human-Computer Interaction (I'm a Dr. now :D)

[Pengfei Zhang @Kindpire]

@Raphael1402X DONE

[Pengfei Zhang @Kindpire]

that reminds me of an episode of Pantheon—the AI 'sees' by playing with sound, and it moves stuff around just by making it vibrate.

[Emanuel Perez @emanperez28]

Astra made me a sonar app that emits undetectable audio to scroll up/down on your computer.

It uses the doppler effect to determine where your hand placement is. You can even double tap in the air to change scroll directions!

[Pengfei Zhang @Kindpire]

@Polymarket Too sensational—just academic research. Social data prediction (e.g., Twitter flu trends) has existed for years; both US & China. The US own Smallville papers pioneered LLM + sociology, with follow-ups simulating consumer/ad feedback and electoral politics. Relax.

[Pengfei Zhang @Kindpire]

@cline I guess K3 is not slow if you deploy it with the same hardware like Fable when it OPEN SOURCE.

[Pengfei Zhang @Kindpire]

No worries, DeepSeek will give us a hundreds-B's version.

[Peter Gostev @petergostev]

Let me show you how to run Kimi-K3 locally, right from your own house

[Unitree @UnitreeRobotics]

Kungfu BOT: Unitree G1🥳
We have continued to upgrade the Unitree G1's algorithm, enabling it to learn and perform virtually any movement. What other moves would you like to see. Do share with us in the comments. (Please keep a safe distance from the robot.)
#Unitree #Kungfu #EmbodiedAI #SpringFestivalGalaRobot #AI #Humanoid #Bipedal #WorldModel #Dance

[SpaceX @SpaceX]

Super Heavy has splashed down in the Gulf of Mexico

[Philipp Schmid @_philschmid]

Is this the architecture behind @OpenAI GPT-4o?
Uni-MoE proposes an MoE-based unified Multimodal Large Language Model (MLLM) that can handle audio, speech, image, text, and video. 👂👄👀💬🎥

Uni-MoE is a native multimodal Mixture of Experts (MoE) architecture with a three-phase training strategy that includes cross-modality alignment, expert activation, and fine-tuning with Low-Rank Adaptation (LoRA). 🤔

TL;DR:
🚀 Uni-MoE uses modality-specific encoders with connectors for a unified multimodal representation.
💡 Utilizes sparse MoE architecture for efficient training and inference
🧑‍🏫 Three-phase training: 1) Train connectors for different modalities 2) Modality-specific expert training with cross-modality instruction data. 3) Fine-tuning with LoRA on mixed multimodal data.
📊 Uni-MoE matches or outperforms other MLLMs on 10 tested vision and audio tasks
🏆 Outperforms existing unified multimodal models on comprehensive benchmarks

Paper: https://t.co/Kupudg0ErG
Github: https://t.co/rAjoF4N2hC

[Elon Musk @elonmusk]

We should send rockets not at each other, but rather to the stars

[Pengfei Zhang @Kindpire]

@ilyasut So what happened, what have you found to force you to make that decision?

[Xingang Pan @XingangP]

Have you thought about interactively 'dragging' objects in the image? Our #SIGGRAPH2023 work #DragGAN makes this come true!🥳

Paper: https://t.co/B3qC0kl1IT
Project page: https://t.co/ZqAEPHNMNF

[AK @_akhaliq]

List of companies working in the AI video space

[AK @_akhaliq]

AI music is taking off, here are some companies working in the space

[Victor M @victormustar]

I just made my own Star Wars clip using AI (text-to-video)!

[Pengfei Zhang @Kindpire]

wow

[aarya @gd3kr]

managed to get the new text to video model running on 12GB of VRAM with half precision. reply here with a prompt and I'll post the generated video

[We Love Movies 🎥 @MoviePolls4U]

On This day in Movie History.

October 28th 1974: JOAQUIN PHOENIX is born.

(🎥 Her - 2013)

[Jackson Greathouse Fall @jacksonfall]

I gave GPT-4 a budget of $100 and told it to make as much money as possible.

I'm acting as its human liaison, buying anything it says to.

Do you think it'll be able to make smart investments and build an online business?

Follow along 👀

[Pengfei Zhang @Kindpire]

@AndrewYNg LLMs are slow and energy-intensive, and many scenarios do not require them. Instead, I guess researchers are building large networks of small, separable sub-networks with millions of parameters for specific ML tasks, reducing computational cost and increasing efficiency.

[Pengfei Zhang @Kindpire]

Please Stop Doing "Explainable" ML - Cynthia Rudin https://t.co/QCNMMZeUfD via @YouTube

[Pengfei Zhang @Kindpire]

lol😂

[Christoph Molnar 🦋 christophmolnar.bsky.social @ChristophMolnar]

when your ml model always predicts the majority class