Dr. Badr M. Abdullah 🇾🇪 (@badr_nlp) 's Twitter Profile
Dr. Badr M. Abdullah 🇾🇪

@badr_nlp

Researcher @LSTSaar | Saarland University🦉 💬 Speech & Language Processing 💫 Machine Learning 🤖 Cognitive Science 🧠

ID: 940954583823765509

linkhttps://badrex.github.io/ calendar_today13-12-2017 14:40:29

1,1K Tweet

936 Takipçi

1,1K Takip Edilen

Dr. Badr M. Abdullah 🇾🇪 (@badr_nlp) 's Twitter Profile Photo

I see a lot of excitement around the "AI Scientist" project But why is this exciting? Who's going to consume/process/evaluate the AI-generated "science"? How will conferences/journals deal with papers generated by "AI scientists"? Where are we going? !!! 🤔

Sarvnaz Karimi (@sarvk) 's Twitter Profile Photo

Really enjoyed this talk "Human Speech Perception in Noise: Can Large Language Models Paraphrase to Improve It?" at HuCLLM ACL 2025 interesting to show how human evaluation is crucial on this task and very well presented too! Anupama Chingacham #ACL2024NLP #NLProc

Really enjoyed this talk "Human Speech Perception in Noise: Can Large Language Models Paraphrase to Improve It?" at HuCLLM <a href="/aclmeeting/">ACL 2025</a> interesting to show how human evaluation is crucial on this task and very well presented too! <a href="/AChingacham/">Anupama Chingacham</a> #ACL2024NLP #NLProc
Alisa Liu (@alisawuffles) 's Twitter Profile Photo

We created SuperBPE🚀, a *superword* tokenizer that includes tokens spanning multiple words. When pretraining at 8B scale, SuperBPE models consistently outperform the BPE baseline on 30 downstream tasks (+8% MMLU), while also being 27% more efficient at inference time.🧵

We created SuperBPE🚀, a *superword* tokenizer that includes tokens spanning multiple words.

When pretraining at 8B scale, SuperBPE models consistently outperform the BPE baseline on 30 downstream tasks (+8% MMLU), while also being 27% more efficient at inference time.🧵
Dr. Badr M. Abdullah 🇾🇪 (@badr_nlp) 's Twitter Profile Photo

This is indeed an interesting approach for Tokenization I wonder how this Tokenizer compares to plain BPE in the presence of character noise, for example, typos and misspellings

Dr. Badr M. Abdullah 🇾🇪 (@badr_nlp) 's Twitter Profile Photo

Fascinating 🤯 Context: humans categorize objects based on global high-level features (e.g., shape), but ImageNet trained vision models exhibit texture bias where low-level features are more important This research revisits the shape vs texture bias with interesting findings

Hosein Mohebbi (@hmohebbi75) 's Twitter Profile Photo

Sadly, I couldn't make #ICML2025 in person to present our work at the exciting Actionable Interpretability Workshop Actionable Interpretability Workshop ICML2025. 📺The recording is online here: youtube.com/watch?v=QkDFCO… +Poster: actionable-interpretability.github.io/posters/poster… 👋Check it out and feel free to get in touch for any questions!

Muhammad AbdulMageed (@mageed) 's Twitter Profile Photo

Join us today at #ACL2025 for the human-centered NLP session. We will be presenting Palm 🏝️, followed by a panel discussion with a lot of timely and engaging talk points. Time: today, Wed July 30 at 9:00-10:30 Place: Hall C, Session 11: IP-Orals, Panel 3: Human-centred NLP

Join us today at #ACL2025 for the human-centered NLP session. We will be presenting Palm 🏝️, followed by a panel discussion with a lot of timely and engaging talk points. 

Time: today, Wed July 30
at 9:00-10:30

Place: Hall C, Session 11: IP-Orals, Panel 3: Human-centred NLP
Dr. Badr M. Abdullah 🇾🇪 (@badr_nlp) 's Twitter Profile Photo

🎤 Just days before #Interspeech2025! We are excited to announce Prof. 𝗞𝗮𝗿𝗲𝗻 𝗟𝗶𝘃𝗲𝘀𝗰𝘂 as our keynote speaker for the special session on Speech Interpretability! 📍 "What can interpretability do for us (and what can it not)?" 📋 More info lnkd.in/eUmR5xbZ

🎤 Just days before #Interspeech2025! 

We are excited to announce Prof. 𝗞𝗮𝗿𝗲𝗻 𝗟𝗶𝘃𝗲𝘀𝗰𝘂 as our keynote speaker for the special session on Speech Interpretability!
📍 "What can interpretability do for us (and what can it not)?" 
📋 More info lnkd.in/eUmR5xbZ
Dr. Badr M. Abdullah 🇾🇪 (@badr_nlp) 's Twitter Profile Photo

If you're in Rotterdam for Interspeech, join us on Monday for the special session on Interpretability in Speech Technology 🕚 11:00 AM 📍Dock 15 🎙️ 𝐊𝐞𝐲𝐧𝐨𝐭𝐞 by 𝑃𝑟𝑜𝑓. 𝐾𝑎𝑟𝑒𝑛 𝐿𝑖𝑣𝑒𝑠𝑐𝑢 🖼️ 𝐏𝐨𝐬𝐭𝐞𝐫 𝐬𝐞𝐬𝐬𝐢𝐨𝐧 for 15 accepted papers #Interspeech2025