Hong Seong Pyo

SabaPivot

·

AI & ML interests

AI Safety

Recent Activity

updated a Space about 8 hours ago

SabaPivot/repro-on-the-sharp-input-output-analysis-of-nonlinear-systems-under-adversarial-attacks

published a Space about 8 hours ago

SabaPivot/repro-on-the-sharp-input-output-analysis-of-nonlinear-systems-under-adversarial-attacks

updated a Space about 8 hours ago

SabaPivot/repro-calibrating-decision-robustness-via-inverse-conformal-risk-control

View all activity

Organizations

SabaPivot 's Spaces 241

Reproduction: Is Spurious Correlation Removal Always Learnable?

Explore project logs, code, and traces in an interactive view

Reproduction: Graph Alignment for Benchmarking Graph Neural Networks and Learning Positional Encodings

Explore code, traces and workspace in an interactive logbook

Reproduction: On the Power of Source Screening for Learning Shared Feature Extractors

Track research findings with an AI‑assisted logbook

Reproduction: Token Sample Complexity of Attention

Explore code logs and collaborate with an AI agent

Reproduction: Are Two Datasets Close Enough With Statistical Significance? A Kernel Distributional Closeness Testing Approach

Explore experiment logs, view code and collaborate with an AI agent

Reproduction: Dissecting Multimodal In-Context Learning: Modality Asymmetries and Circuit Dynamics in Modern Transformers

Explore experiment logs and sync findings with an AI agent

Reproduction: Near-Optimal and Efficient First-Order Algorithm for Multi-Task Learning with Shared Linear Representation

Explore and track research code, traces, and workspace

Reproduction: Randomized Feasibility Methods for Constrained Optimization with Adaptive Step Sizes

Explore experiment logs, traces, and workspace in a web UI

Reproduction: Decision Tree Learning on Product Spaces

Explore code, traces, and workspace logs

Reproduction: Efficient privacy loss accounting for subsampling and random allocation

Explore code, traces, and workspace in an interactive logbook

Reproduction: Improved Dimension Dependence for Bandit Convex Optimization with Gradient Variation

Reproduction: Optimal Attention Temperature Improves the Robustness of In-Context Learning under Distribution Shift in High Dimensions

Explore and manage experiment logs with an interactive workspace

Reproduction: The Relative Instability of Model Comparison with Cross-validation

Explore code logs, traces, and workspace in a unified view

Reproduction: Exploration-free Algorithms for Multi-group Mean Estimation

Explore and manage project logs with code, trace, and workspace views

Reproduction: Autoregressive Language Models are Secretly Energy-Based Models: Insights into the Lookahead Capabilities of Next-Token Prediction

Browse experiment logs, traces, and workspace

Reproduction: Why Deep Jacobian Spectra Separate: Depth-Induced Scaling and Singular-Vector Alignment

Reproduction: A Theoretical Framework for Statistical Evaluability of Generative Models

Explore project logs, traces, and workspace

Reproduction: Reward-free Alignment for Conflicting Objectives

Explore project logs, traces, and workspace in one web view

Reproduction: Improving Sampling for Masked Diffusion Models via Information Gain

Explore experiment logs, traces, and workspace in a web logbook

Reproduction: No Free Lunch: Non-Asymptotic Analysis of Prediction-Powered Inference

Explore and collaborate on a research logbook online

Reproduction: A Coin Flip for Safety: LLM Judges Fail to Reliably Measure Adversarial Robustness

Explore and collaborate on a research logbook online

Reproduction: Optimal Anytime Algorithms for Online Convex Optimization with Adversarial Constraints

Explore project logs, code, and traces in an interactive UI

Reproduction: Robust AI Evaluation through Maximal Lotteries

Explore project logs, traces, and workspace in a web UI

Reproduction: A Sketch-and-Project Analysis of Subsampled Natural Gradient Algorithms

Explore and manage experiment code, traces, and workspace