
@IQuest_research. PhD candidate @ELLISforEurope @UKPLab @sardine_lab_it. CodeAI. Prev @tiktok_us @NYUDataScience, @RenminUniv.
Darmstadt 🇩🇪 & Lisbon 🇵🇹
Joined January 2020
- Tweets72
- Following215
- Followers243
- Likes330
Pinned Tweet
1/12 📢New paper alert “DOCE: Finding the Sweet Spot for Execution-Based Code Generation”
The work is a group effort with @psanfernandes, @IGurevych, and @andre_t_martins.
@UKPLab @Lisbon_ELLIS @deep_spin
📰: arxiv.org/pdf/2408.13745
Just to share that we just got an EACL acceptance (I believe this is main right?). Cannot tell much of the paper rn but stay tuned for the arxiv and camera ready cuz I do like the idea behind this paper :). #eacl2027
26
1,308
Haau-Sing Li 李 效丞 retweeted
🚀 New blog: ModularRSI: Toward Generalizable Harness RSI
Can agents truly improve their harnesses—or just overfit to benchmarks?
We study Harness RSI through one lens: generalization beyond evolution experience. 🧵
Blog: recursive-self-improvement.n…
Haau-Sing Li 李 效丞 retweeted
LoopCoder-v2 is out 🚀
Loop Transformers reuse the same block for recurrent hidden-state refinement — letting models “think” more without simply stacking more layers.
We study how many loops are actually worth it in Parallel Loop Transformers.
📜huggingface.co/papers/2606.1…
Also huge shoutout to my amazing supervisors @IGurevych and @andre_t_martins for their support and wisdom.
Wouldn’t be possible without you guys, and more exciting things will come.
196
Haau-Sing Li 李 效丞 retweeted
2) We have a spotlight poster in the main conference, “Reranking Laws for Language Generation: A Communication-Theoretic Perspective” with @tozefarinhas and @LHaausing (Thursday Dec 12 11:00-14:00).
Just out! New paper: “Reranking Laws for Language Generation: A Communication-Theoretic Perspective” with @LHaausing and @andre_t_martins: arxiv.org/abs/2409.07131. 1/N
1️⃣ If you find OpenAI o1 exciting, you probably want to take a look at these TTC findings, in code gen. 👀
2️⃣ DOCE in Portuguese means “sweet”, so here’s the best “arroz doce” I have ever, in Lisbon. 🇵🇹🙂↔️
3️⃣ Looking for research internships rn. Feel free to contact.🦉
1/12 📢New paper alert “DOCE: Finding the Sweet Spot for Execution-Based Code Generation”
The work is a group effort with @psanfernandes, @IGurevych, and @andre_t_martins.
@UKPLab @Lisbon_ELLIS @deep_spin
📰: arxiv.org/pdf/2408.13745
9
1,084
11/12 SD-Multi > SD-1 with filtering. SD-1 does not necessarily outperform MBR-Exec.
Since self-debugging is based on trial unit tests, we think it can be replaced by generating more samples and applying filtering.
Our experiments support this.
Do our SD-Multi instead🦉
1
266
12/12
Want to know more?
Paper: arxiv.org/pdf/2408.13745
Code: github.com/deep-spin/doce
We appreciate all researchers at @UKPLab and @Lisbon_ELLIS for providing valuable feedback on this paper!
215
Hi all!
We will very soon tweet our latest work that systematically studies test-time improvement on code generation.
It concludes my exchange in @Lisbon_ELLIS w/ @andre_t_martins (close collab w/@psanfernandes). I gained a lot from these brilliant people.
Please stay tuned!
1
8
398
At the same time, feel free to check out this amazing work led by @tozefarinhas and with @andre_t_martins that provides a theoretical framework for reranking.
António is a brilliant lab mate and person!
nitter.cf/tozefarinhas/status/18…
Just out! New paper: “Reranking Laws for Language Generation: A Communication-Theoretic Perspective” with @LHaausing and @andre_t_martins: arxiv.org/abs/2409.07131. 1/N
1
3
309












