Life update: Thrilled to share that I'll be joining
@UCLAComSci
as an Assistant Professor in 2025-26 academic year!
Our lab will focus on Efficient AI and Quantum Computing. We have multiple openings, more info at
Hinton helped me realize my childhood dream—to become a physicist. 🙂
There’s no doubt about Hinton’s contributions to humanity, but with the slowing progress in fundamental science, is it appropriate to award the most representative prize in fundamental science to the field of
“How could I be sure it wasn’t a spoof call?”
2024 physics laureate Geoffrey Hinton received a phone call from Stockholm in the early hours in a hotel room in California. Multiple Swedish accents helped reassure him that his
#NobelPrize
in Physics, awarded today, was real.
Very humbled to receive the 1st place of ACM Student Research Competition for my work on ML for Quantum and Quantum for ML! Very thankful to my Advisor Prof.
@SongHan_MIT
, all my mentors and collaborators! Also thank Prof. Callie Hao, Prof. Debjit Pal for organizing!
@ICCAD
Great experience attending ISSCC 2024 at SF and excited to see many ICs for quantum technology!🔥 Also it's my honor to receive the 𝐑𝐢𝐬𝐢𝐧𝐠 𝐒𝐭𝐚𝐫 𝐀𝐰𝐚𝐫𝐝 𝐨𝐟 𝐈𝐒𝐒𝐂𝐂 2024 for our work on "SpAtten" Sparse Attention Accelerator. 1/3
Avi Wigderson’s lecture on “Reading Alan Turing’s Papers”. Interesting to see a Turing Award winner focus not on his own work, but on the foundations of TCS through Turing’s lens. Last point of his summary on Turing is “Does not rush to publish” :)
HPCA 2022 "QuantumNAS" covered by MIT News. We propose an efficient method to search for noise-robust circuit architecture (Ansatz) of parameterized quantum circuits.
TorchQuantum open-source:
Website:
Grateful to receive the Nvidia Graduate Fellowship Finalist and the present a RTX 3090! very heavy but super cool! Thanks
@nvidia
!
(Anyone interested in donating another 7? so I will have a 8-GPU machine😆)
It's a great honor to receive the Best Pitch Award at MIT Microsystems Annual Research Conference 2023! I learnt a lot from the diverse formats and eye-opening creativity of over 100 research pitches from the MTL and MIT.nano community.
#quantum
#MTL
@microsystechlab
@mit_nano
Grateful to receive the
@Qualcomm
Innovation Fellowship 2021! Thank my advisor
@SongHan_MIT
, Professor Hae-Seung Lee, and collaborators for the warm supports! Will continue exciting work on efficient Machine Learning + quantum computing.
I'm attending the 59th DAC and will be presenting 2 papers:
"QuantumNAT: Quantum Noise-Aware Training with Noise Injection, Quantization and Normalization" on July 12, 10:30 PDT; and "QOC: Quantum On-Chip Training with Parameter Shift and Gradient Pruning
" on July 13, 13:30 PDT.
Appreciate the opportunity to teach DNN training at MIT course "Hardware Architecture for Deep Learning" developed by Prof. Joel Emer and Prof. Vivienne Sze.
Wonderful experience at IEEE QCE (Quantum Week) 2023! Excited to see the rapid growth of the community!
P1. P2. Glad to receive the Best Paper Award, thanks to all my advisors and collaborators for the support, especially to Han Zheng and
@junyuliu4
for leading the efforts!
First-author paper "HAT: Hardware-Aware Transformers for Efficient Neural Machine Translation" accepted by
#ACL2020
. We use NAS to search different small&fast models for different hardware in an enlarged design space. Thank my advisor and all collaborators!
#acl2020
#acl2020nlp
Welcome to attend the Quantum Computer Systems lecture series!
Sign-up:
Website:
We are also welcoming new speakers to share your research, please contact hanrui
@mit
.edu and zliang5
@nd
.edu
#quantumcomputing
#systems
#quantum
Welcome to attend Lecture 1 of the Quantum Computer Systems Series! We are fortunate to have Prof. Yongshan Ding at Yale talking about quantum noise mitigation🙌
Time: Jun 30, 10:30 AM ET
Zoom:
#quantum
#computing
#system
#lecture
Glad to be selected as a Rising Star in ML and Systems! Hope to discuss and learn more from other attendees of the workshop. Many thanks to my advisor
@songhan_mit
and collaborators for the support!
#ml
#systems
#rising_star
We are pleased to present the inaugural MLCommons Rising Stars cohort. This talented group of PhD students are the future leaders of ML and Systems research.
Happy to share our new paper accepted to
@HpcaArchConf
2021. We propose efficient algorithm-architecture codesign for Attention in NLP by removing computations and DRAM accesses of unimportant tokens & heads of the model in a cascade manner. Thanks my advisor and collaborators!
Our SpAtten NLP accelerator is selected as MIT Homepage Spotlight! Thanks my advisor
@SongHan_MIT
and collaborators for the warm support!
More info on efficient NLP at
I will be attending 𝐐𝐂𝐄 (𝐐𝐮𝐚𝐧𝐭𝐮𝐦 𝐖𝐞𝐞𝐤) in Broomfield CO! Very excited to share our recent work!
𝐓𝐨𝐫𝐜𝐡𝐐𝐮𝐚𝐧𝐭𝐮𝐦 𝐓𝐮𝐭𝐨𝐫𝐢𝐚𝐥
𝐕𝐚𝐫𝐢𝐚𝐭𝐢𝐨𝐧𝐚𝐥 𝐏𝐮𝐥𝐬𝐞 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠
#QCE
#quantum
#tutorial
I'm incredibly grateful to my PhD advisor
@songhan_mit
, for his unwavering support in both research and life. I'd also like to extend my heartfelt thanks to my mentors Anantha Chandrakasan,
@prof_chong
, Jason Cong,
@creamyoki
, Harry Lee, David Z. Pan, Richard Shi and many others
DAC 2020 first-author paper accepted! We applied RL and GNN for better transistor sizing. Thank my advisor
@SongHan_MIT
and collaborators for the kind support!😀
#57DAC
@57thDAC
NLP's Moore's Law: every year model size increases by 10x!😇
How to make them efficient? Check out our work on efficient NLP🙌
Hardware-Aware Transformer:
NLP Attention accelerator:
Prune/Quantize LMs:
JMLR paper: MicroNet for Efficient Language Modeling. Won the 1st at MicroNet challenge NeurIPS'19
arXiv:
talk (26:17):
code:
project page:
Happy playing with the language models!
Our
#acl2020nlp
"HAT: Hardware-Aware Transformers" is open-sourced!
Paper:
Website:
We search a fast model for each hardware in a supernet with latency feedback.
Pull request and ⭐ are welcomed!
#NLP
#NLProc
My last class of "Hardware Architecture for Deep Learning" as Teaching Assistant this semester. Memorable online TA experience, and made good friends with students.🙌
As well as my collaborators, family, and friends, for their invaluable support throughout this amazing journey! Super excited to begin the new chapter!
I will do a gap year based in Boston. Come say hi when you are around!
#quantum
#ai
#ml
#mlsys
@UCLA
@UCLAengineering
QOC proposes to train the parameterized quantum circuits on quantum machines to improve the scalability and prune unreliable gradients to reduce cost. Paper:
Attending
@ICCAD
2022 at San Diego, will present:
1. Student Research Competition titled "ML for Quantum and Quantum for ML".
Website:
Poster:
2. TorchQuantum tutorial
3. Top3 team at TinyML Contest
#quantum
I will be giving a talk on "SpArch: Efficient Architecture for Sparse Matrix Multiplication" in HPCA 2020. See you today 17:00 @ Capri/Rivera
Project:
Lightning Talk:
#hpca2020
@HpcaArchConf
Gave a talk on "TTS: Transferable Transistor Sizing with Graph Neural Networks and Reinforcement Learning" at Qualcomm Research Center. Thank Joseph for hosting, many inspiring discussions!
Coming to
@ICCAD
@ SF. Presenting some preliminary results on reliable quantum state preparation and QEC decoding. Also attending the award ceremony of quantum computing for drug discovery challenge and present our solution. 🚀
#quantum
#qec
#architecture
sad... I respect many pioneering work ZAPATA has done, but commercializing quantum AI is challenging at this stage due to the small size, low reliability of existing quantum hardware
$ZPTA big news according to its new filings today:
<Summary of Key Points>
-Costs Associated with Exit or Disposal Activities (Item 2.05): On October 7, 2024, Zapata Computing's board approved the termination of all employees, except a few retained
"AI for EDA", promising or not? Check out our
#57dac
paper "GCN-RL Circuit Designer: Transferable Transistor Sizing With Graph Neural Networks and Reinforcement Learning"!
Paper:
Video:
Website:
I will be at the online Q&A for our
#57dac
paper "GCN-RL Circuit Designer" tomorrow (Friday) afternoon
Time: 5pm EDT to 6:30pm EDT
@57thDAC
session: "Learning to yield"
Zoom link:
Welcome!
Project website:
It’s hard to cool something that’s already very cold.
As we scale towards useful fault-tolerant quantum computers, the iconic “chandelier”-style cryostats often used in lab experiments are struggling with the increasing heat load associated with more complex chips and wiring.
Finish a fun summer internship
@nvidia
Architecture Research Group. Big thank to my great mentors
@Joel_Emer
, Mike, Donghyuk and Steve! Two big differences for research in industry (1) much easier tech transfer (2) communication (meetings) takes a larger part of your time.
#NVIP
Paper sharing "ConfuciuX: Autonomous Hardware Resource Assignment for DNN Accelerators using RL"
Idea: Given a DNN model&dataflow, using RL to search for a best hardware resources allocation strategy.
1. Why only target at spatial PE based arch?
@ying11231
I think one is 1956 Nobel Prize in Physics, to William Shockley, John Bardeen and Walter Brattain for discovery of the transistor effect; another is 2000 Nobel Prize in Physics, to Jack Kilby, Zhores Alferov and Herbert Kroemer, for inventing integrated circuits