Somin W
@SominW
Followers
260
Following
2K
Statuses
127
RT @ChantalShaib: headed to miami/#EMNLP2024 next week with this work 🌴🌴 come check out the poster (12/11) and let’s talk about templates i…
0
7
0
RT @sheridan_feucht: Presenting this short paper at #EMNLP2024 main conference next Tuesday, with a bonus @BlackboxNLP poster on Friday! 🥃…
0
7
0
Planning to attend @emnlpmeeting? Come check out our work/poster in Miami on 11/12 🏝️
📢We know that including CoT rationales as supervision improves model distillation. But why? New work (w/ @silvio_amir and @byron_c_wallace) research unveils surprising insights! 🔗 Full paper: [1/5]
1
4
11
RT @ChantalShaib: There's a general feeling that AI-written text is repetitive. But this repetition goes beyond phrases like "delve into"!…
0
110
0
RT @sheridan_feucht: (new preprint) LLMs live in a strange tokenized world. We find that LLMs learn to deal with the weirdness of tokenizat…
0
60
0
4⃣Our work opens new avenues for efficient model distillation, reducing the need for extensive reasoning sequences and focusing on critical token signals. Dive into the details and explore our ablations![5/5] 🔗Read more:
0
0
0
@kuchhal_dhruv It’s not everywhere because a significant majority of people hate it. It decreases consumer trust since most people do not have the flexibility of showing up at “off-peak” hours, and therefore incur price increase. Beyond that, price stability is highly desirable.
0
0
1
RT @monicamreddy: 1/10 Excited to share our #NAACL2024 paper “On-the-fly Definition Augmentation of LLMs for Biomedical NER”, paper 📄 : htt…
0
9
0
RT @RohitGandikota: Have you ever wanted to make a precise change when generating images with diffusion models?🎨 We present Concept Sliders…
0
37
0
RT @kpal_koyena: #LLMs are trained to predict the next word. But we found that they are overachievers! When GPT-J is busy predicting the n…
0
52
0
@krismicinski Some version (varying proportions) of this -- (depending on who you broker with). Increase bond % as you age etc. US treasuries are also cool rn with near-guaranteed ~5% return, at least for the next year or so. ETF equivalent could be SGOV/USFR/BIL.
0
0
2
RT @ChantalShaib: 🤔 Thinking about using GPT to summarize medical evidence? Check out our #ACL2023NLP short where we evaluate GPT-3's abil…
0
14
0
👉 Check out the full paper for more details on how larger language models can push the limits of generative approaches to RE and yield competitive performance in supervised learning with (very) limited data.
0
0
0