Zhemal Khamidun

ETH and Roche Train AI to Predict Where C–H Borylation Will Occur on a Molecule

PoS Method Catches AI Agents Stuck in Loops of False Belief: 36 Repeats in ALFWorld

Robotics

Ego Data Gets a "Full Body": A New Scaling Path for World Models

Business·Google DeepMind Blog

Google DeepMind and Singapore Release Gemma-Based Scam Detector as Fraud Losses Top S$4 Billion

GovTech and Google DeepMind fine-tuned Gemma on local scam patterns. The model runs on a smartphone, reads chats without the cloud, and warns users before they send…

SAE-IDF Matches Symbols Across Different Systems and Beats a Baseline by Up to 45 Points

SAE-IDF uses sparse autoencoders to match elements across heterogeneous symbolic systems. It beats a dense baseline on 12 tasks with Gemma-2-9B, winning by up to 45…

ReaDy speeds up hybrid LLM decoding 3.16x on a single B200

ReaDy feeds the recurrent states of hybrid linear models into a speculative drafter, averaging a 3.16x speedup on Qwen3.8-27B and a single B200.

DAGFormer Plans Connections Between Layers and Cuts Perplexity by 17–25%

DAGFormer adds direct cross-layer connections to the transformer and plans them with a topology predictor. On models up to 1B, WikiText-2 perplexity drops by 17–25%.

01Researchers Find a 0.4–1.1% Window in Pythia Pretraining Where the Readout ReorganizesOct 14, 2026·2 min 02semCER Weights OCR Errors by Severity: 61M LLM-Judge RatingsOct 14, 2026·2 min 03SAE on Gemma-2-2B: Share of Matched Features Rises from 2% to 29% With 50M TokensOct 14, 2026·2 min 04REASON teaches a model to think in two loops and saves about half the computeOct 14, 2026·2 min 05AA-Robotics v0.1: GPT-6 Astra controls a robot arm and scores 69 out of 100Oct 14, 2026·3 min

Top themes

MULTIFINGROUND: LLMs Recognize Accounting Concepts With 99% Accuracy but Miss Facts

Agent teams for $155 instead of $510: Cohere finds where delegation actually pays off

Cohere ran a controlled study of how to organize teams of coding agents. The finding: a leader on the Sol model paired with 16 cheap Luna agents for planning…

Robotics·Hugging Face Daily Papers

WinARDet fixes looping in MLLM detectors: F1 rises from 32.16 to 39.36

The WinAR window-compression method stops multimodal models from repeating one box endlessly: on the dense RSDense benchmark, F1 rose to 39.36 and duplicates fell…

GAFU pulls knowledge out of generation in UMMs and lifts VStar-Bench by 20%

The GAFU post-training method compresses a unified multimodal model's generative representations into G-tokens and gains up to +20.00% over SFT on VStar-Bench.

RTAformer Teaches a Model to Notice When Relationships Between Objects Change in Video

MolLean Makes LLMs Check Their Chemistry: Wins on 12 of 12 Tasks

PrismaCoder: Open 9B and 35B Models Close In on Proprietary Ones in Visual Code Generation

FLOMSS: flow matching splits music into 4 stems and takes 2nd place in listening tests

SliderEdit++ Smooths Out Sliders in AI Image Editors: +18.5% and Up to 20x Faster

JockeyBench: 28 LLM Agent Configurations All Lost Money on Japanese Horse Racing

KANLab: A TensorFlow Playground Fork That Shows KAN Training in the Browser