Search result for:
events & Conferences
Snorkel AI
at NeurIPS 2025
From benchmark innovation to hands-on workshops, our team will be on-site to share how data-centric methods are advancing model reliability, evaluation, and agent intelligence.
San Diego Convention Center
December 2-7, 2025
Our featured research papers
We’re proud to share several papers accepted to NeurIPS 2025—spanning LLM evaluation, agent learning environments, and advances in multi-modal and theoretical ML.
RESEARCH PAPER
Shrinking the generation-verification gap with weak verifiers
Weaver combines multiple weak verifiers into a single strong one, improving model accuracy without heavy labeled data or compute.
RESEARCH PAPER
From many voices to one: Statistically principled aggregation of LLM judges
We introduce CARE, a confounder-aware LLM-as-a-judge aggregation framework that explicitly models latent biases and reduces aggregation error.
RESEARCH PAPER
Beyond Accuracy: Dissecting LLM mathematical reasoning under RL
Introducing SPARKLE, a fine-grained framework that reveals how RL shapes LLM reasoning across planning, knowledge integration, and subproblem solving.
RESEARCH PAPER
Theoretical Physics Benchmark—a dataset & study of AI reasoning capabilities in theoretical physics
We introduce a benchmark to evaluate the capability of AI to solve problems in theoretical physics.
RESEARCH PAPER
LLM-integrated Bayesian state space models for multi-modal time-series forecasting
We integrate LLMs with a Bayesian state space model to jointly perform numeric and textual forecasting.
Featured workshops
Our accepted research papers will be featured across several workshops.
Featured WORKSHOP
Sunday, December 7, 2025
Scaling Environments for Agents (SEA) Workshop
Snorkel is proud to sponsor the SEA Workshop, joining leading researchers from Stanford, Anthropic, DeepMind, and other frontier labs to advance work on intelligent agents powered by large language models (LLMs).
The workshop highlights how scalable, diverse, and high-fidelity environments are as essential to agent intelligence as data and compute are to model performance.
Featured panelist
Frederic Sala
Chief scientist, Snorkel AI
WORKSHOP
Saturday, December 6, 2025
Reliable ML from Unreliable Data
This workshop bridges theory and practice to tackle these challenges, bringing together researchers working on distribution shift, adversarial robustness, and strategic behavior to chart principled yet deployable solutions.
WORKSHOP
Sunday, December 7, 2025
Evaluating the Evolving LLM Lifecycle
As LLMs rapidly integrate into diverse applications, the pressing challenge is not just to evaluate their current performance, but to define the next generation of evaluation protocols for increasingly capable and complex LLMs. This workshop addresses this need for robust methods and best practices across the entire LLM lifecycle.
WORKSHOP
Sunday, December 7, 2025
Machine Learning and the Physical Sciences (ML4PS)
This year's programming explores the evolving interplay between academia and industry in basic research. Invited talks and panel discussions emphasize the myriad foundational and translational connections between these domains.
WORKSHOP
Sunday, December 7, 2025
Recent Advances in Time Series Foundation Models (BERT2S)
This workshop aims to bring together researchers to examine the gap between TSFM potential and real-world utility, and to identify benchmarks and applications where TSFMs can truly excel.
Meet our researchers on-site
From benchmark innovation to hands-on workshops, our team will be on-site to share how data-centric methods are advancing model reliability, evaluation, and agent intelligence.
Frederic Sala
Chief scientist
Snorkel AI
Justin Bauer
Research engineer
Snorkel AI
Derek Pham
Research engineer
Snorkel AI
Jason Qi
Research scientist
Snorkel AI
