Jia-Bin Huang
@jbhuang0604
Followers
65K
Following
8K
Media
2K
Statuses
5K
Expectation of CS PhD applicants over the years. 🫤 Have experiences doing research. ☺️ Publish a top conference paper. 😮 Pass 1,000 citations on GScholar. 🤯 Solve a moonshot project that blows everybody’s mind in a three-month research visit.
11
18
515
RT @derekhalpern: PSA: Evernote raised their monthly fee of my account from $65.14 to $1361.83. I have received monthly receipts from them….
0
174
0
A new proposal from @CMSgov under @DrOz wants to add catheters, ostomy & trach supplies to Medicare’s competitive bidding program. This is a bad idea and will worsen outcomes for seniors & Medicare recipients. My latest in @MedPageToday explains why ⬇️
medpagetoday.com
Limiting urological, tracheostomy, and ostomy supply options will do more harm than good
9
12
115
New video! . A quick dive into the recent Hierarchical Reasoning Model (HRM) through the lens of algorithm synthesis. Check it out:
7
64
554
Teaching (multimodal) foundation models this coming semester!. The field has moved so fast and it’s so hard to keep up! Any pointers and resources on interesting topics? Eg foundation models for robotics, world model, agentic AI, RL for LLMs, efficient attention, long-context.
12
16
328
What was I thinking in my single-author article .
arxiv.org
Recent years have witnessed a significant increase in the number of paper submissions to computer vision conferences. The sheer volume of paper submissions and the insufficient number of competent...
to really understand a scholar's thinking, read their single-author articles.
1
1
50
Just watched the Demon Slayer infinity castle… and Holy Crap!. I know it would be good but didn’t know it would be this good. My jaw was on the floor the whole time. The visuals, the story, the sound, the music, everything was breathtaking!
2
0
54
I think the story is like…. We want to train the model to minimize the KL divergence from predicted distribution to the ground truth one. Basically measures the extra bits we wasted because we use the wrong distribution. (See . BUT, the GT distribution.
Why is cross-entropy a good loss for language pretraining? . I do think it is, but I am curious what is the commonly accepted intuition behind it.
4
10
199
For those interested in learning more, here are some (incomplete) references. Continuous diffusion on word/paragraph embeddings: .- Diffusion-LM: .- LD4LG: .- PLANNER: - SLD:
arxiv.org
Diffusion models have shown promise in text generation, but often struggle with generating long, coherent, and contextually accurate text. Token-level diffusion doesn't model word-order...
2
1
28
Diffusion LLMs are promising ways to overcome the limitations of autoregressive LLMs. Less error propagation, easier to control, and faster to sample! . But how do Diffusion LLMs actually work? 🤔. Let's explore some ideas on this fascinating topic!
18
109
769
RT @KalshiSports: Game by the numbers. Kalshi volume: $26.6m.Spit ejections: 1.Rizzler commercials: 1.Weather delays: 1.AJ Brown catches: 1.
0
34
0
Academic paper rebuttal POV:. R2: “Thanks for addressing all my concerns. I will maintain my score as borderline reject.”
7
9
269
Woohoo! Imagine, Verify, Execute (IVE) is accepted to CoRL 2025! 🎉. Congrats to the incredible @umdcs students Seungjae Lee @JayLEE_0301, Daniel Ekpo (@daniekpo7), Haowen Liu!.
Exploration is key for robots to generalize, especially in open-ended environments with vague goals and sparse rewards. BUT, how do we go beyond random poking? Wouldn't it be great to have a robot that explores an environment just like a kid?. Introducing Imagine, Verify,
1
9
58
This is a short clip from the full video on MeanFlow. Check it out if you are interested in one-step generative models.
0
6
35
Who is Adam? . In this video, meet this guy who has some momentum and his cousin AdamW.
3
11
146
Decoding comments on academic talks. Great talk: You did okay. Nice talk: I was checking my phone most of the time. Love the intro: I stopped paying attention after slide 3. Thanks for your talk: You SUCK!.
1
3
39
Bitcoin’s on fire at $112K! Time to flip the charts on BTCC!.Exploring Cryptocurrency with Jaren Jackson Jr.🏀.
0
1
9
In an era of billion-parameter models everywhere, it's incredibly refreshing to see how a fundamental question can be formulated and solved with simple, beautiful math. - How should we orient a solar panel ☀️🔋? -. Zero AI! If you enjoy math, you'll love this!
3
27
278
*Slides without slide titles*. When I first tried presenting WITHOUT slide titles, everything flowed so much better! (totally validated . by me)! . Give it a shot! Once you try it, you’ll never want to go back.
4
0
20