Sahil Kadadekar

Crusadersk

https://chimeraforge.vercel.app/reports

AI & ML interests

LLM inference, quantization, refusal robustness, model evaluation, GPU serving, batching effects, post-training, MLOps, AI safety. Reports: https://chimeraforge.vercel.app/reports

Recent Activity

updated a Space 3 days ago

Crusadersk/icml26-row-stochastic-decentralized-repro

published a Space 3 days ago

Crusadersk/icml26-row-stochastic-decentralized-repro

updated a Space 3 days ago

Crusadersk/icml26-llm-as-judge-scaling-repro

View all activity

Organizations

Crusadersk 's Spaces 48

Repro - Row-Stochastic Matrices Can Provably Outperform Doubly Stochastic Matrices in Decentralized Learning

🎯

Explore research logbooks and sync findings with an AI agent

Repro - Demystifying LLM-as-a-Judge: Inference-Time Scaling

🎯

Collaborate on a shared experiment logbook with an AI agent

Repro - Rex: A Family of Reversible Exponential (Stochastic) Runge-Kutta Solvers

🔁

Track and collaborate on experiment logs with an AI agent

Repro - Maximum Likelihood Reinforcement Learning

🎲

Explore RL experiment logs and sync with your coding agent

Repro - A Random Matrix Perspective on the Consistency of Diffusion Models

🎲

Explore a research logbook and collaborate with an AI agent

Repro - Diffusion Bridge or Flow Matching? A Unifying Framework

🌉

Browse an interactive research logbook on diffusion models

Repro - NeuralFLoC: Neural Flow-Based Joint Registration and Clustering of Functional Data

🌊

Explore experiment logs and sync with an AI agent

Repro - Tucker Attention: A generalization of approximate attention mechanisms

🧩

Track experiments and sync findings with an AI agent

Repro - Explaining Concept Shift with Interpretable Feature Attribution

🔎

Explore and collaborate on a research logbook with AI assistance

Repro - Optimal Unconstrained Self-Distillation in Ridge Regression

🎯

Explore research logbooks and collaborate with an AI agent

Repro - Quantifying LLM Attention-Head Stability: Implications for Circuit Universality

🧭

Explore experiment logs and sync findings with an AI agent

Repro - DAVE: Distribution-Aware Attribution via ViT Gradient Decomposition

🔍

Explore and edit experiment logs with AI collaboration

Repro - Foundations of Equivariant Deep Learning

🔺

Browse experiment logbook and sync findings with your coding agent

Repro - Trees to Flows and Back: Unifying Decision Trees and Diffusion Models

🌲

Explore experiment logs and sync findings with an AI agent

Repro - High-accuracy sampling for diffusion models and log-concave distributions

🎯

Explore experiment logs and sync them with a coding agent

Repro - Nonparametric Distribution Regression Re-calibration

🎯

Browse experiment logs and collaborate with an AI agent

Repro - Parametrized Power-Iteration Clustering for Directed Graphs

🧭

Explore experiment logbook and sync findings with an AI agent

Repro - On Structured State-Space Duality

🎯

Browse and sync project logbook with an AI agent

Repro - Improved Analysis of the Accelerated Noisy Power Method with Applications to Decentralized PCA

🌀

Browse and collaborate on research logbooks online

Repro - Attention's forward pass and Frank-Wolfe

🧭

View and collaborate on project logs with an AI agent

Repro - Minimum Distance Summaries for Robust Neural Posterior Estimation

🎯

Explore experiment logs and sync findings with an AI agent

Repro - Beyond Softmax: A Natural Parameterization for Categorical Random Variables

🎯

Browse and collaborate on research logbooks

Repro - DropoutTS: Sample-Adaptive Dropout for Robust Time Series Forecasting

🎯

Explore project logs and sync findings with an AI agent

Repro - The Theory and Practice of MAP Inference over Non-Convex Constraints

📐

Explore and share experiment logs with an AI collaborator

Sahil Kadadekar

AI & ML interests

Recent Activity

Organizations

Crusadersk 's Spaces 48 Sort: Recently updated

Repro - Row-Stochastic Matrices Can Provably Outperform Doubly Stochastic Matrices in Decentralized Learning

Repro - Demystifying LLM-as-a-Judge: Inference-Time Scaling

Repro - Rex: A Family of Reversible Exponential (Stochastic) Runge-Kutta Solvers

Repro - Maximum Likelihood Reinforcement Learning

Repro - A Random Matrix Perspective on the Consistency of Diffusion Models

Repro - Diffusion Bridge or Flow Matching? A Unifying Framework

Repro - NeuralFLoC: Neural Flow-Based Joint Registration and Clustering of Functional Data

Repro - Tucker Attention: A generalization of approximate attention mechanisms

Repro - Explaining Concept Shift with Interpretable Feature Attribution

Repro - Optimal Unconstrained Self-Distillation in Ridge Regression

Repro - Quantifying LLM Attention-Head Stability: Implications for Circuit Universality

Repro - DAVE: Distribution-Aware Attribution via ViT Gradient Decomposition

Repro - Foundations of Equivariant Deep Learning

Repro - Trees to Flows and Back: Unifying Decision Trees and Diffusion Models

Repro - High-accuracy sampling for diffusion models and log-concave distributions

Repro - Nonparametric Distribution Regression Re-calibration

Repro - Parametrized Power-Iteration Clustering for Directed Graphs

Repro - On Structured State-Space Duality

Repro - Improved Analysis of the Accelerated Noisy Power Method with Applications to Decentralized PCA

Repro - Attention's forward pass and Frank-Wolfe

Repro - Minimum Distance Summaries for Robust Neural Posterior Estimation

Repro - Beyond Softmax: A Natural Parameterization for Categorical Random Variables

Repro - DropoutTS: Sample-Adaptive Dropout for Robust Time Series Forecasting

Repro - The Theory and Practice of MAP Inference over Non-Convex Constraints

Crusadersk 's Spaces 48