Yi Zhang
@Wado_Will
Followers
177
Following
685
Media
1
Statuses
108
Researcher & Founder ｜Ph.D. @Penn ❤️ Boxing 🥊 & Tennis🎾
NYC
Joined September 2016
1. This is not the academic convention for coauthorship in CS. 2. If you operate by this convention, consider whether there are more fulfilling ways you could be spending your time.
9
12
250
From https://t.co/YF0ytRqpKN: “On Thursday, Greenshoe, a generative artificial intelligence-powered platform designed to help draft U.S. Securities and Exchange Commission (SEC) disclosures, announced a $3 million seed funding round, as well as the expansion of its platform’s
law.com
Law.com delivers news, insights and resources that allow legal professionals to anticipate opportunities, adapt to change, and prepare for future success.
0
1
0
A rough structure for telling the story better would be: - Problem - Why it matters - What others missed - The key idea - Evidence - Limitations - Takeaway
0
2
33
The reviewer who wrote the negative review often did not participate in the discussion at all.
The NeurIPS acceptance bar is very high, and papers with a negative score rarely get accepted. Even when a negative review is poorly written—sometimes clearly generated by ChatGPT—it can still strongly influence the final decision, since the AC must keep the acceptance rate.
0
0
0
New Zuck post, what a difference a few years makes: Today: "We'll need to be rigorous about mitigating these risks and careful about what we choose to open source." 2024: "Meta is committed to open source AI... and therefore a platform that will be around for the long term."
56
64
696
GitHub Copilot is an interesting exercise in the idea that no matter how good a thing is, there is always a newer thing that could be significantly better. Once you internalize this the world around you is just so obviously full of potential.
61
92
2K
As a researcher, it's easy to get distracted by what others are working on. I've seen many people conducting research on problems they don't genuinely care about—just because the community values them (e.g., solving Math Olympiad problems). It's important to focus on research
18
42
441
🤔
let's talk about "agents" (in the LLM sense). there's a lot of buzz around "multi-agent" systems where agents collaborate but... i don't really get how it differs from a thinking of a single agent with multiple modes of operation. what are the benefits of modeling as multi-agent?
0
0
0
Attending 🌴#EMNLP2024 or interested in what people are working on these days? We organized it all for you with Knowledge Navigator! Explore all @emnlpmeeting accepted papers mapped by themes and subtopics—giving you a bird’s-eye view of the conference https://t.co/Jom6WEmZzc
3
23
78
New SIGMOD blog post: "Where does database research go from here?" where I examine the academic database research community from the lens of market competition and ask " what is our competitive advantage?" https://t.co/7oKwyN3ljy
8
36
178
We’re excited to share our #ACL2024 conference and findings papers in our newest blog post: https://t.co/rYS6ZkP7hV. Check out these papers from today’s sessions and next week’s virtual Findings presentations! Congrats to @keviv9
@Wado_Will
@DanRothNLP and their co-authors!
0
4
7
If you are a student or academic researcher and want to make progress towards human-level AI: >>>DO NOT WORK ON LLMs<<< LLMs are an off ramp. Thousands of engineers are working on LLMs with enormous computing resources. The only way you could possibly contribute is by analyzing
Great talk by @ylecun yesterday, at the scientific symposium for the opening of the @ELLISInst_Tue! I took the liberty of summarizing one of his main take-home messages...
290
945
6K
It's a great article to read! I'd like to add one other crucial cost in the calculation: people cost. The conventional cloud market is relatively heavy in software. AWS is called "web service" and not "Amazon Cloud" because there are a lot of manpower to build all the software
GPU Cloud Economics Explained The Hidden Truth CPU vs GPU Cloud Differences, TCO Model, PUE, Hyperscalers Disadvantage https://t.co/5b3H6jXbJf
5
19
52
"Not working on XML in the early 2000s was like not working on LLMs right now." - anonymous professor
2
1
12
Notable: while LLMs are the singular force shaping the discipline & dominating the discourse in NLP, I saw absolutely nobody from the major LLM shops (@OpenAI, @AnthropicAI, @MosaicML) make an appearance at #acl2023.
19
26
342
LM APIs are fickle, hurting reproducibility (I was really hoping that text-davinci-003 was going to stick around for a while, given the number of papers using it). Researchers should seriously use open models (especially as they are getting better now!) https://t.co/z8MS5QvEom
GPT-4 API is now available to all paying OpenAI API customers. GPT-3.5 Turbo, DALL·E, and Whisper APIs are also now generally available, and we’re announcing a deprecation plan for some of our older models, which will retire beginning of 2024:
7
42
258
I think it would be in all of our interests if the whole AI community came together to fix this.
11
5
49
We are hiring science interns for summer 2023, here is the link to apply
1
3
5
ACL/ICML highlights threats of LMs like ChatGPT generating paper content. However, I'm more concerned about reviews. Example: ChatGPT generates confident but nonsensical reviews better than any R2. Now, it's AC's job to spot real reviews, but I lack confidence in doing so
4
22
203