It's quite odd that the models from my ACL work has been downloaded over 3,000 times on Hugging Face but has only been cited once. 🥲 I might have to advertise the work more.

After ACL2023, our 'FACTUAL' model for text scene graph parsing, a joint effort with Wuhan University and Adobe Research, exceeded 4,000 downloads on Hugging Face! 😂 I just spent a weekend updating it for easier use. Check it out: [github.com/zhuang-li/Fact…].

I am extremely honoured and pleased to share our survey titled: Natural Language Processing for Dialects: A Survey. arxiv.org/abs/2401.05632 Ours is the first of its kind, comprehensive survey of NLP for dialects. Aditya Joshi Diptesh Kanojia Reza Haffari

How ICL 𝘦𝘮𝘦𝘳𝘨𝘦𝘴 from unsupervised data? 𝘐𝘵 𝘭𝘦𝘢𝘳𝘯𝘴 𝘧𝘳𝘰𝘮 parallel phrases After deleting parallel parts the ICL ability was reduced by 51% deleting random words - only 2% 🧵 Yanda Chen Chen Zhao zhou Yu He He Columbia NLP arxiv.org/abs/2402.12530

Introducing: StarCoder2 and The Stack v2 ⭐️ StarCoder2 is trained with a 16k token context and repo-level information for 4T+ tokens. All built on The Stack v2 - the largest code dataset with 900B+ tokens. All code, data and models are fully open! hf.co/bigcode/starco…

Long (code) files may not be as high quality as you think… Excited for our new work, "Brevity is the soul of wit: Pruning long files for code generation". We find that long code files are often low quality and show benefits from pruning such files for code gen. Read on 🔎⏬

My friend mentioned his cat always crawls on his keyboard when he is coding, so here's my solution for mine. 😂

Full benchmarks for LLaMA-3.1 (both instruct and base):

who knows what is the board line score for ACL rolling review… I thought it was 3 but now I am a little bit confused …

I found 2 out of 4 ARR reviews were generated with Claude. I can even reproduce their reviews with my Claude account… can I complain to the chairs. But their scores are actually good to me

BTW. I found the LLM review tends to add a short title with a paragraph of long explanation. if anyone uses LLMs to review the paper, please at least modify the writing style. Otherwise it is too obvious…