Yunmo Chen

71 posts

Yunmo Chen

Yunmo Chen

@YunmoChen

MTS @MicrosoftAI | Previously @Bloomberg @jhuclsp @Apple @MSFTResearch @Amazon | Opinions are my own

New York, USA Katılım Temmuz 2017
486 Takip Edilen377 Takipçiler
Yunmo Chen retweetledi
Yu Su
Yu Su@ysu_nlp·
Introducing @NeoCognition, the agent lab for specialized intelligence. Everyone needs experts, but human expertise does not scale. Backed by $40M seed funding, we build self-learning agents that specialize across domains to make expertise abundant.
English
92
134
852
162.1K
Yunmo Chen
Yunmo Chen@YunmoChen·
🌐 Check it out If you’re interested in robust, adaptive retrieval mechanisms and RL applications in NLP, don’t miss our paper! Feedback and discussions welcome! 📜 Link: aclanthology.org/2024.emnlp-mai…
English
1
0
0
203
Yunmo Chen
Yunmo Chen@YunmoChen·
Excited to share our latest work, Learning to Retrieve Iteratively for In-Context Learning, at #EMNLP2024! Here’s how we’re pushing the boundaries of retrieval techniques in large language models (LLMs). 🧵👇
English
1
5
15
4.6K
Yunmo Chen retweetledi
Haoran Xu
Haoran Xu@fe1ixxu·
Here’s some better news: Combining CPO and SimPO can likely improve the model! Check out more details in our GitHub code: github.com/fe1ixxu/CPO_SI…
Haoran Xu tweet media
Haoran Xu@fe1ixxu

We recently had multiple rounds of discussions with the SimPO authors regarding the lack of comparison to CPO in their main paper. We both agree that it was an unintentional oversight, and they will update the paper to address it. We appreciate their positive and prompt response in engaging with us to resolve the issue! After my personal experience and reproduction of SimPO, I would like to give a shout out to SimPO—an impressively stable reference-free preference learning method compared to others!

English
1
14
37
12.3K
Yunmo Chen retweetledi
Weiting (Steven) Tan
Weiting (Steven) Tan@weiting_nlp·
Will be presenting my work on LLM-MT at #naacl today, at In-Person Poster Session 7 (11am). Please come if you are interested in machien translation and multilinguality! Would also love to discuss multimodal LLM (speech + text) and non-autoregressive decoding (NAT, diffusion)!
Weiting (Steven) Tan@weiting_nlp

📚 LLMs benefit from demonstrations, but can we make zero-shot translations better without access to parallel corpora? Find out how we bridge the gap between zero- and few-shot translation with only monolingual information in our latest paper! 🤓 arxiv.org/abs/2311.02310 @jhuclsp

English
0
5
15
2.2K
Yunmo Chen
Yunmo Chen@YunmoChen·
@yc6n13 @fe1ixxu For policy-based learning, our experience is that we need something that regularizes the space so that the policy won’t deviate too much from its original behavior while gradually adopting to desired preferences. This is the reason that in our CPO we added a BC loss.
English
0
0
4
475
Tony Yu Cao
Tony Yu Cao@yc6n13·
@fe1ixxu From most data points I have collected, SimPO doesn’t work as expected. There are too many similar xPO claim improvements over DPO but lack of good verification & validation.
English
3
0
9
1.6K

Keşfet

@NeoCognition @chenwanch1 @kesnet50 @ctongfei @harsh_jhamtani @nlpaxia @adveisner @ben_vandurme