Guy Bar-Shalom

[1/7] New paper: "Beyond Token Probes: Hallucination Detection via Activation Tensors with ACT-ViT" #NeurIPS2025 [arxiv.org/pdf/2510.00296] Joint work with: @ffabffrasca (co-first), @yaniv_galron, @YftahZ , @HaggaiMaron

Check out our new ICLR 2026 paper - we explore hallucination detection through graph learning. Take a look!

These works were joint efforts with a group of amazing collaborators: @ffabffrasca @HaggaiMaron @YftahZ @ytn_ym @yaniv_galron @yoav_gelberg @itayevron @mayabechlerspei @ido_guy Ami Tavory Moshe Eliasof Ran Elbaz

📌 [4/4] On the Expressive Power of GNN Derivatives We study how using gradients of GNNs can increase their expressive power, providing a principled way to go beyond standard message passing. arxiv.org/pdf/2510.02565

Happy to share my new #ICLR2026 papers !

📄 Beyond Token Probes: Hallucination Detection via Activation Tensors with ACT-ViT w/ @GuyBarSh , @ffabffrasca, Yaniv Galron, @YftahZ arxiv.org/abs/2510.00296

@GuyBarSh and I will be presenting the poster today, stop by 🤗 📍 Fri, Dec 5 • 4:30–7:30 PM PST • Exhibit Hall C,D,E # 4000

🤔 Can discrete diffusion models actually outperform standard classifiers? We show that it can! 📄 arxiv.org/pdf/2511.20263 💻 github.com/omerb01/didicm 🌐 omerb01.github.io/didicm-web

[7/7] Code: github.com/BarSGuy/ACT-ViT

[6/7] Results (over 15 LLM/dataset combinations): • Consistently outperforms classic probes • Zero-shot generalization to new datasets • Fast adaptation to unseen LLMs by tuning only their new corresponding adapter

[1/7] New paper: "Beyond Token Probes: Hallucination Detection via Activation Tensors with ACT-ViT" #NeurIPS2025 [arxiv.org/pdf/2510.00296] Joint work with: @ffabffrasca (co-first), @yaniv_galron, @YftahZ , @HaggaiMaron

“Balancing Efficiency and Expressiveness: Subgraph GNNs with Walk-Based Centrality” is at #ICML2025! Drop by our poster on Tuesday 11:00–13:30, East Exhib. Hall A-B # E-3005. Joint work w/ @JoshSouthern13 @ytn_ym @GuyBarSh @mmbronstein @HaggaiMaron arxiv.org/abs/2501.03113 🧵

We believe LOS-Net is a step toward transparent, and scalable (gray-box) LLM analysis. Inspired by works by: @WeijiaShi2 , @anirudhajith42 , @LukeZettlemoyer , @OrgadHadas , @boknilev and more. 🔗 Paper: [arxiv.org/pdf/2503.14043] 💻 Code: [github.com/BarSGuy/LLM-Ou…] #AI #LLMs

📈 Results Highlights: • SOTA performance on Hallucination & Data Contamination Detection (DCD) • Strong cross-LLM and cross-dataset generalization See the figure for zero-shot cross-LLM generalization on the BookMIA DCD benchmark (ROC-AUC):

📢 Introducing: Learning on LLM Output Signatures for Gray-box LLM Behavior Analysis [arxiv.org/pdf/2503.14043] A joint work with @ffabffrasca (co-first author) and our amazing collaborators: @dereklim_lzh @yoav_gelberg @YftahZ @el_yaniv @GalChechik @HaggaiMaron 🧵Thread