Images are great way to understand AI slop visually, this is a random frame of this popular video. There are three microphones, the two directional microphones aren’t even pointing at her and one seems to be pointing behind her (this is absurd) The camera behind seems to be

To define slop we need to formalize what an “attention economy” really is. When you (or an LLM) reads something, you invest attention. There’s a cost economically and in time. You pay it because you’re hoping to get something out of reading. Some spark of insight, some deeper

"Measuring How (Not Just Whether) Vision-Language Models Build Common Ground" We propose a new metric suite to evaluate how VLMs build common ground in dialogue, not just describe images. 🧵👇

🚀 Excited to share our new paper: “SiLVERScore: Semantically-Aware Embeddings for Sign Language Generation Evaluation”, presented at RANLP 2025. 🧵 A thread on why current metrics fail and how SiLVERScore changes that.

I would like to thank my wonderful co-authors, Anthony Sicilia and @malihealikhani Only with their efforts and constant support, this paper could come together. 🧵n/n

Code and checkpoints are open-source! ⚠️These LLMs are not perfect, and we will be updating as new libraries become available. Dive in and build with us the next generation of signing LLMs: 🔗 https://t.co/qqcFxGhfEq 🔗 https://t.co/GQ4gEjkqLG 🧵7/n

🛠️ In the paper, we also explored all avenues to teach LLMs to sign! We tested: 🌸In-Context Learning (Prompt-Tuning) 🌸Supervised Fine-Tuning (SFT) 🌸Multitasking Fine-Tuning 🧵6/n

🛡️ Forget Forgetting! We solve the problem of catastrophic forgetting. When an LLM loses its original spoken language skills after learning a new task. Our Multitasking Fine-Tuning strategy (mixing DGS & spoken language data, OpenOrca) successfully mitigates this data shift.🧵5/n

🚀 Our Core Contribution: We introduced new fine-tuning strategies to build the first text-based and multimodal LLMs capable of SLP. This includes: ✅ Video-Based Input capabilities via fine-tuned LLaVA (V2T task). ✅ Text-Based Models fine-tuned on LLaMA3. 🧵4/n

Unlike other LLMs, our models are trained and tested on 6 additional SLP tasks: 🌿(G2T) DGS to German 🌿(T2G) German to DGS 🌿(V2T) DGS Videos to German 🌿(I-G2T) Intensified DGS to German 🌿(T2I-G) German to Intensified DGS 🌿(G2E) DGS to English 🧵 3/n

🚀 We're proud to be among the first to bring multimodal and text-based SLP models to the open-source community. Our approach is uniquely comprehensive. 🧵 2/n

🏃‍♀️💨While waiting for SignGemma to become available... You can check out our first text-based and multimodal LLMs capable of Sign Language Processing (SLP) called SignAlignLM! #SignLanguage #LLM #AIAccessibility @aclmeeting 📜Paper: https://t.co/BVAFmhNzlw 🧵1/n

📣 Excited to announce SpaVLE: #NeurIPS2025 Workshop on Space in Vision, Language, and Embodied AI! 👉 https://t.co/teKhSn5Qo0 🦾Co-organized with an incredible team → @fredahshi · @maojiayuan · @DJiafei · @ManlingLi_ · David Hsu · @Kordjamshidi 🌌 Why Space & SpaVLE? We

Türkiye'de Eylül 2024'te başlatılan, Mart 2025 ile iyice netleşen rejim değişikliği girişimini @cakir_rusen'le konuştuk. Başlığı sadece sosyal bilimcileri ilgilendiriyormuş gibi gözükse de hepimizin geleceğini etkileyecek olan bu meseleyi merak edenler 👇 https://t.co/DxBNQxL36X

Özgür Özel tepkilerden çekinmedi, LeMan’a sahip çıktı: “Ben bakınca Gazze’de hayatını kaybetmiş bir melek görüyorum. ‘Peygamberlerin adını alanlar ölüyor’ diye resmedilmiş bir karikatür ama kolay! Saldırın LeMan’a! O LeMan, hepiniz susarken Mavi Marmara’ya destek karikatürü

🚨Reminder: Submissioms for the ORIGen workshop at COLM are due today!!! 🚨 CfP: https://t.co/WnHx62WjiJ OpenReview submission page:

People often claim they know when ChatGPT wrote something, but are they as accurate as they think? Turns out that while general population is unreliable, those who frequently use ChatGPT for writing tasks can spot even "humanized" AI-generated text with near-perfect accuracy 🎯

new multi-turn instruction grounding dataset with @wp_mccarthy and @saujasv - multi-modal instruction : drawing + txt - verifiable execution : 2D CAD gym env - easy eval : API → score - baselines : human vs VLMs - large : 15,163 inst-exe rounds https://t.co/CxehBadcgu [1/n]

The AI boom means the world’s data centers use more electricity than almost every country. See our blog on how to supply the power-hungry technology that’s driving growth. https://t.co/sgGEpNsGmY

Really excited to share our #ICLR2025 paper! We propose ACT, an approach for improving LLMs’ ability to navigate multi-turn conversations via implicit action optimization. Our LLMs are better equipped to achieve multi-turn goals and recognize when to ask clarifying questions!