neurips-2026

[NeurIPS 2026] VLA-Corrector: Lightweight Detect-and-Correct Inference for Adaptive Action Horizon

StepShield: When, Not Whether to Intervene on Rogue Agents — NeurIPS 2026 benchmark for temporal evaluation of AI agent guardrails (9,429 trajectories, step-level annotations)

[NeurIPS 2026] StableHand: Quality-Aware Flow Matching for World-Space Dual-Hand Motion Estimation from Egocentric Video

[NeurIPS 2026] FlowHOI: Flow-based Semantics-Grounded Generation of Hand-Object Interactions for Dexterous Robot Manipulation

[NeurIPS 2026] ReasoningShield: Safety Detection over Reasoning Traces of Large Reasoning Models

NeurIPS 2026: Mechanistic Interpretability toolkit for Vision-Language-Action models

K-Veritas CLI

The Neural Interface Foundation Challenge — NeurIPS 2026 Sydney. An open-source brain-decoding benchmark with four tracks across EEG, EMG, sleep, and BCI.

Code, models and data for M-plicits: Neural Implicit Surfaces via Nested Multiscale Residuals (NeurIPS 2026). A noise-robust neural surface representation built on nested neighborhood bands, with real-time multiscale sphere tracing and analytical normals.

interactive, sub-16ms vector explainer built for the DataForge 2026 Pathway Track.It proves why classical State Space Models suffer exponential associative forgetting ($\rho(A)^L$) and visualizes how Pathway's Dragon Hatchling (BDH) synaptic plasticity and BDH CQ continuous latent reasoning eliminate interference without KV-cache explosion.

[NeurIPS 2026] Official implementation for "The Dormant Spiking Neuron: A State-Driven Mechanism for Efficient Spiking Neural Networks"

[NeurIPS 2026] Linguistic Trajectory Encoding (LTE): object-centric spatiotemporal memory for embodied agents. CPU and real-video demos with SAM3, ViPE, DINOv2, and Qwen3-VL. Spatial Memory Benchmark (SMB). arXiv:2609.04802.

Submission for RoCo-Spring (NeurIPS 2026) — Robust Correspondence Challenge on Optical Flow, Stereo Matching, Scene Flow, and Exploration tracks using the Spring/RobustSpring benchmark.

🔥 🔥 [NeurIPS 2026] "Decoupled Mode Connectivity for Base-to-Novel Generalization in Vision-Language Models"

Phase transitions in AI scaling. The alignment tax is engineerable. 63 base + 39 frontier models. arXiv: 2605.18838, 2605.18840

Official website for the NeurIPS 2026 Workshop on AI for Chip Design.

CALM: interpretable by design — accurate locally additive models for tabular data (NeurIPS 2026)

QuantAlpha: Deterministic Credit Assignment in Multi-Agent Financial Systems via Causal Provenance Ledgers (ACM SIGKDD 2026 FedMAS)

CarbonBench: A Knowledge-Graph Benchmark for Retrieval-Augmented QA over Carbon Credit Documents (NeurIPS 2026 Datasets and Benchmarks Track, under review)

SSIT: a label-free, gold-free test for whether LLM-judge position x verbosity bias corrections actually compose. No human labels, no model of the judge. Code, 7-judge/6-family pilot data, and pre-registered protocol for the NeurIPS 2026 JUDGe workshop paper.