KeepCurrent @_keepcurrent

We organize #meetups, #hackathons and conduct #tailormade #trainings and #workshops for companies: #MachineLearning #MLOps, #DataScience & #SoftwareDevelopment. keep-current.com Vienna, Austria Joined May 2018

Tweets

114
Followers

95
Following

296
Likes

365

Hila Gonen @hila_gonen

a year ago

Do you like yellow? Then, according to LLMs, you are probably a school bus driver! Excited to share our new paper about Semantic Leakage in Language Models! Joint work with my wonderful collaborators @Terra @alisawuffles @luke @nlpnoah Paper: gonenhila.github.io/files/Semantic… 1/10

11 34 204 31K 64

Download Image

Vaibhav (VB) Srivastav @reach_vb

a year ago

Probably the craziest week in Open Source AI (yet): 1. Mistral (in collaboration with Nvidia) dropped Apache 2.0 licensed NeMo 12B LLM, better than L3 8B and Gemma 2 9B. Models are multilingual with 128K context and a highly efficient tokenizer - tekken. 2. Apple released DCLM…

4 137 734 108K 647

Tiago Pimentel @tpimentelms

2 years ago

Are you interested in word lengths and natural language’s efficiency? If yes, our new #EMNLP2023 paper has everything you need: drama, suspense, a new derivation of Zipf’s law, an update to Piantadosi et al’s classic word length paper, transformers... 🧵 arxiv.org/abs/2312.03897

1 23 185 35K 75

Download Image

Zhiqiu (Oscar) Xu @oscar_zhiqiu_xu

2 years ago

You don’t have to train from scratch whenever developing a smaller model of an existing model family. Sharing our latest work - “Initializing Models with Larger Ones” arxiv preprint: arxiv.org/abs/2311.18823 code: github.com/OscarXZQ/weigh…

5 51 349 50K 221

Download Image

Aviv Slobodkin @lovodkin93

2 years ago

🎉Excited to announce our paper's acceptance at #EMNLP2023! We explore a fascinating question: When faced with (un)answerable queries, do LLMs actually grasp the concept of (un)answerability?🧐 This work is a collaborative effort with @clu_avi @ravfogel @omerNLP and Ido Dagan 1/n

2 26 96 18K 39

Samuel Müller @SamuelMullr

2 years ago

There is a paper by Google trending right now, that claims transformer in-context learning cannot generalize between two function classes I have reproduced their experiment in a colab and come to a very different conclusion...

12 136 1K 379K 886

Download Image

ACL 2025 @aclmeeting

2 years ago

ACL org announcement: 📢The list of accepted workshops in the ACL Conferences (@aclmeeting, @eaclmeeting, @naaclmeeting, @emnlpmeeting) in 2024 is out! Please help spread the word. Retweeting w/ references, esp. w/organisers information is very much appreciated - thanks! #NLProc

4 56 148 32K 57

Bojan Tunguz @tunguz

2 years ago

Pandas 2.0 is here! This is the biggest overhaul of Pandas since its inception, and it has been years in the making. However, you will probably not notice too many changes, and all your existing Pandas code will most likely run the same as before. All the major changes are under…

10 152 950 158K 198

Download Image

Steve Stewart-Williams @SteveStuWill

3 years ago

Psychologists have posited hundreds of cognitive biases over the years. A new paper argues that they all boil down to one of a handful of fundamental beliefs coupled with confirmation bias. doi.org/10.1177/174569…

89 2K 8K 2.0M 5K

Download Image

Guillaume Lample @ NeurIPS 2024 @GuillaumeLample

3 years ago

Today we release LLaMA, 4 foundation models ranging from 7B to 65B parameters. LLaMA-13B outperforms OPT and GPT-3 175B on most benchmarks. LLaMA-65B is competitive with Chinchilla 70B and PaLM 540B. The weights for all models are open and available at research.facebook.com/publications/l… 1/n

158 1K 7K 3.2M 2K

Download Image

DeepAI @DeepAI

3 years ago

Read about Non-Parametric Model deepai.org/machine-learni… #OrdinalNumber #NonParametricModel

0 2 5 941 0

NeurIPS Conference @NeurIPSConf

3 years ago

You can now watch the recorded material from #NeurIPS2022 online without registration at: slideslive.com/neurips-2022

8 218 780 140K 261

Science Is Strategic @scienceisstrat1

3 years ago

1/ The 2022 climate data is out. And it’s terrifying. Overall, 2022 was the fifth warmest year on record. Below is a short🧵on 2022’s record-breaking droughts, floods, ‘rain bombs,’ wildfires & more Cc: @dwallacewells @JesseJenkins @ylecun @leahstokes @Noahpinion @MichaelEMann

56 542 989 368K 165

Download Image

Sasha Rush @srush_nlp

3 years ago

If you are looking for a winter break project, here is the full collection of ML/coding puzzles. (I think this is more useful than prompting, but who knows?) * github.com/srush/tensor-p… * github.com/srush/gpu-puzz… * github.com/srush/autodiff… * github.com/srush/raspy

23 106 824 145K 649

Download Image

Simone Scardapane @s_scardapane

3 years ago

*Thinking Like Transformers* Awesome blog post by @srush_nlp based on the paper by the same name. If you write a programming language inspired by the way Transformers work, how easy would it be to program in it? 👀 Blog: srush.github.io/raspy/ Paper: arxiv.org/pdf/2106.06981…

6 131 608 70K 272

Download Image

Fangyu Liu @hardy_qr

3 years ago

📍🧵🚨 QA on plots & charts is a complex task requiring sophisticated reasoning - our visual language models struggle with this. LLMs are super strong reasoners - but they only work for text. What do we do? We translate plots & charts to text so LLM can understand!

3 16 84 11K 19

Download Image

Abhilasha Ravichander @lasha_nlp

3 years ago

🚨Help NLP models process negation🚨 Introducing ℂ𝕆ℕ𝔻𝔸ℚ𝔸, a *contrastive* reading comprehension dataset that requires reasoning about negation w/ @nlpmattg & @anmarasovic @ai2_allennlp, at #EMNLP2022 📝Paper arxiv.org/abs/2211.00295 🚀Data github.com/AbhilashaRavic… [1/8]

8 50 254 0 60

Download Image

Aman Jha @amanjha__

3 years ago

It's here! Upload *any paper* to Explainpaper and start instantly getting explanations! Ask follow up questions if you need a more in-depth answer. Go to explainpaper.com and go read all the papers you've been saving! 📝📝📝

160 1K 5K 0 3K

Download Video

Mathias Lechner @mlech26l

3 years ago

We (joint work with @ramin_m_h) have released PyHopper, a hyperparameter tuning platform for streamlining machine learning research. Pyhopper's goal is to enable people of any skill level to set up advanced multi-GPU hyperparameter tuning processes in less than a minute.

1 2 20 0 6

Download Image

GenBench @GenBench

3 years ago

Ever wanted to know more about generalisation in NLP but overwhelmed with the number of papers on ArXiv? Fear not! We read 400+ papers, 600+ experiments, and designed a taxonomy 📝 to categorise the research for you! (1/n) 🧵 arxiv.org/abs/2210.03050