Ben Bogin (@ben_bogin) 's Twitter Profile
Ben Bogin

@ben_bogin

CS PhD student at Tel-Aviv University, studying #NLProc.
benbogin.github.io

ID: 150610839

calendar_today01-06-2010 10:56:41

136 Tweet

640 Followers

428 Following

Aizenberg (@aizenberg55) 's Twitter Profile Photo

THREAD: Gaza as “open prison” or “caged” is a core lie by Israel haters & anti-Zionists. Part of thinking is to never give credit for Israel leaving Gaza permanently as it shatters entire narrative of “perpetual occupation”; all blame MUST remain on Israel. Here's the truth 1/

Hen Mazzig (@henmazzig) 's Twitter Profile Photo

Hamas: “we will repeat the October 7 massacre time and again, 1M times if we need to, until we end the occupation.“ Journalist: “occupation of Gaza?” Hamas: “no, all of Israel.”

Maayan Zin (@zinmaayan1007) 's Twitter Profile Photo

Worldwide people chant for "Palestine to be Free" yet my two daughters, captives of Hamas, are ignored. Even as thousands in London push for ceasefire, there's silence on their release. I'm heartbroken by the world's indifference. Dafna and Ella, I miss you so much 💔

Worldwide people chant for "Palestine to be Free" yet my two daughters, captives of Hamas, are ignored. Even as thousands in London push for ceasefire, there's silence on their release. I'm heartbroken by the world's indifference.

Dafna and Ella, I miss you so much 💔
Ai2 (@allen_ai) 's Twitter Profile Photo

OLMo is here! And it’s 100% open. It’s a state-of-the-art LLM and we are releasing it with all pre-training data and code. Let’s get to work on understanding the science behind LLMs. Learn more about the framework and how to access it here: blog.allenai.org/olmo-open-lang…

AK (@_akhaliq) 's Twitter Profile Photo

Allen AI presents Dolma an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research paper page: huggingface.co/papers/2402.00… dataset: huggingface.co/datasets/allen… release Dolma, a three trillion tokens English corpus, built from a diverse mixture of web content,

Allen AI presents Dolma

an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research

paper page: huggingface.co/papers/2402.00…

dataset: huggingface.co/datasets/allen…

release Dolma, a three trillion tokens English corpus, built from a diverse mixture of web content,
Haidar Khan (@haidarkk1) 's Twitter Profile Photo

Skeptical about LLM benchmarks telling the whole story? 🤔 Tiny tweaks to tests like MMLU can shuffle model rankings like a deck of cards. 🃏 Our latest work delves into #LLM benchmarks to highlight this ArXiv link: arxiv.org/abs/2402.01781

Skeptical about LLM benchmarks telling the whole story? 🤔 Tiny tweaks to tests like MMLU can shuffle model rankings like a deck of cards. 🃏
  
Our latest work delves into #LLM benchmarks to highlight this 
ArXiv link: arxiv.org/abs/2402.01781
Maor Ivgi (@maorivg) 's Twitter Profile Photo

1/5 🧠 Excited to share our latest paper focusing on the heart of LLM training: data curation! We train a 7B LLM achieving 64% on 5-shot MMLU, using only 2.6T tokens. The key to this performance? Exceptional data curation. #LLM #DataCuration

Maor Ivgi (@maorivg) 's Twitter Profile Photo

1/7 🚨 What do LLMs do when they are uncertain? We found that the stronger the LLM, the more it hallucinates and the less it loops! This pattern extends to sampling methods and instruction tuning. 🧵👇 Mor Geva Jonathan Berant Ori Yoran

1/7 🚨 What do LLMs do when they are uncertain? We found that the stronger the LLM, the more it hallucinates and the less it loops! This pattern extends to sampling methods and instruction tuning. 🧵👇
<a href="/megamor2/">Mor Geva</a> <a href="/JonathanBerant/">Jonathan Berant</a> <a href="/OriYoran/">Ori Yoran</a>