author
Chief Scientist
,
Snorkel AI
Assistant Professor @ University of Wisconsin-Madison
Frederic Sala is Chief Scientist at Snorkel AI and an assistant professor in the Computer Sciences Department at the University of Wisconsin-Madison. His research studies the fundamentals of data-driven systems and machine learning, with a focus on data-centric AI, foundation models, and automated machine learning. He and his group received the 2024 DARPA Young Faculty Award, a best student paper runner-up award at UAI ’22, the outstanding Ph.D. dissertation award from the UCLA Department of Electrical Engineering, the NSF Graduate Research Fellowship.
The latest from Fred
LiveResearchBench: A Live Benchmark for User-Centric Deep Research in the Wild
Deep research---producing comprehensive, citation-backed reports by searching across hundreds of live websites---marks an important frontier for agentic systems. To rigorously evaluate this ability, three principles are essential: tasks should be (1) user-centric, reflecting realistic information needs, (2) dynamic, requiring up-to-date information beyond parametric knowledge, and (3) unambiguous, ensuring consistent interpretation across users. Existing benchmarks fall short of these principles, often focusing on narrow domains or posing ambiguous questions that hinder fair comparison. Guided by these principles, we introduce LiveResearchBench, a benchmark of 100 expert-curated tasks spanning daily life, enterprise, and academia, each requiring extensive, dynamic, real-time web search and synthesis....
Research Paper
Accepted to ICLR 2026
LiveResearchBench: A Live Benchmark for User-Centric Deep Research in the Wild
Deep research—producing comprehensive, citation-backed reports by searching across hundreds of live websites—marks an important frontier for agentic systems. To rigorously evaluate this ability, three principles are essential: tasks should be (1) user-centric, reflecting realistic information needs, (2) dynamic, requiring up-to-date information beyond parametric knowledge, and (3) unambiguous, ensuring consistent interpretation across users. Existing benchmarks fall short of these principles, often…
Personalize Your LLM: Fake it then Align It
Personalizing large language models (LLMs) is essential for delivering tailored interactions that improve user experience.Many existing personalization methods require fine-tuning LLMs for each user, rendering them prohibitively expensive for widespread adoption.Although retrieval-based approaches offer a more compute-efficient alternative, they still depend on large, high-quality datasets that are not consistently available for all users.To address this challenge, we propose CHAMELEON, a scalable and efficient personalization approach that uses (1) self-generated personal preference data and (2) representation editing to enable quick and cost-effective personalization.Our experiments on various tasks, including those from the LaMP personalization benchmark, show that CHAMELEON efficiently adapts models to...
Research Paper
Accepted to NAACL Findings 2025
Personalize Your LLM: Fake it then Align It
Personalizing large language models (LLMs) is essential for delivering tailored interactions that improve user experience.Many existing personalization methods require fine-tuning LLMs for each user, rendering them prohibitively expensive for widespread adoption.Although retrieval-based approaches offer a more compute-efficient alternative, they still depend on large, high-quality datasets that are not consistently available for all users.To address this challenge, we propose CHAMELEON, a…
Test-Time Scaling Makes Overtraining Compute-Optimal
Modern LLMs scale at test-time, e.g. via repeated sampling, where inference cost grows with model size and the number of samples. This creates a trade-off that pretraining scaling laws, such as Chinchilla, do not address. We present Train-to-Test (T2) scaling laws that jointly optimize model size, training tokens, and number of inference samples under fixed end-to-end budgets. T2 modernizes pretraining scaling laws with pass@k modeling used for test-time scaling, then jointly optimizes pretraining and test-time decisions. Forecasts from T2 are robust over distinct modeling approaches: measuring joint scaling effect on the task loss and modeling impact on task accuracy. Across eight downstream tasks, we find...
Research Paper
Accepted to COLM
Test-Time Scaling Makes Overtraining Compute-Optimal
Modern LLMs scale at test-time, e.g. via repeated sampling, where inference cost grows with model size and the number of samples. This creates a trade-off that pretraining scaling laws, such as Chinchilla, do not address. We present Train-to-Test (T2) scaling laws that jointly optimize model size, training tokens, and number of inference samples under fixed end-to-end budgets. T2 modernizes pretraining scaling laws…
OR-LLM-Bench: A Pipeline for Scalable and Verifiable Text-to-Optimization Synthesis
Operations research (OR)-style modeling poses challenges for large language models (LLMs). It requires long-context consistency, producing precise mathematical formulations, and the ability to infer implicit constraints. To study these challenges under controlled conditions, we build a verifiable synthetic pipeline that generates large-scale certified optimization problem instances. Using this pipeline, we obtain several insights: first, direct natural language translation of optimization problems runs into an \emph{effective context limit}, beyond which frontier models abruptly fail to maintain global variable–constraint consistency---despite remaining within nominal context window length. Second, naive divide-and-conquer scaling strategies struggle due to context explosion and semantic fragmentation. Third, while frontier...
Research Paper
Accepted to ICLR 2026
OR-LLM-Bench: A Pipeline for Scalable and Verifiable Text-to-Optimization Synthesis
Operations research (OR)-style modeling poses challenges for large language models (LLMs). It requires long-context consistency, producing precise mathematical formulations, and the ability to infer implicit constraints. To study these challenges under controlled conditions, we build a verifiable synthetic pipeline that generates large-scale certified optimization problem instances. Using this pipeline, we obtain several insights: first, direct natural language translation of optimization…
Expressivity-Efficiency Tradeoffs for Hybrid Sequence Models
Hybrid sequence models--combining Transformer and state-space model layers--seek to gain the expressive versatility of attention as well as the computational efficiency of state-space model layers. Despite burgeoning interest in hybrid models, we lack a basic understanding of the settings where--and underlying mechanisms through which--they offer benefits over their constituent models. In this paper, we study this question, focusing on a broad family of core synthetic tasks. For this family of tasks, we prove the existence of fundamental limitations for non-hybrid models. Specifically, any Transformer or state-space model that solves the underlying task requires either a large number of parameters or...
Research Paper
Accepted to ICML 2026
Expressivity-Efficiency Tradeoffs for Hybrid Sequence Models
Hybrid sequence models–combining Transformer and state-space model layers–seek to gain the expressive versatility of attention as well as the computational efficiency of state-space model layers. Despite burgeoning interest in hybrid models, we lack a basic understanding of the settings where–and underlying mechanisms through which–they offer benefits over their constituent models. In this paper, we study this question, focusing on a…
Weight Updates as Activation Shifts: A Principled Framework for Steering
Activation steering promises to be an extremely parameter-efficient form of adaptation, but its effectiveness depends on critical design choices -- such as intervention location and parameterization -- that currently rely on empirical heuristics rather than a principled foundation. We establish a first-order equivalence between activation-space interventions and weight-space updates, deriving the conditions under which activation steering can replicate fine-tuning behavior. This equivalence yields a principled framework for steering design and identifies the post-block output as a theoretically-backed and highly expressive intervention site. We further explain why certain intervention locations outperform others and show that weight updates and activation updates play...
Research Paper
Accepted to ICML 2026
Weight Updates as Activation Shifts: A Principled Framework for Steering
Activation steering promises to be an extremely parameter-efficient form of adaptation, but its effectiveness depends on critical design choices — such as intervention location and parameterization — that currently rely on empirical heuristics rather than a principled foundation. We establish a first-order equivalence between activation-space interventions and weight-space updates, deriving the conditions under which activation steering can replicate fine-tuning behavior….
RoboCritics: Enabling Reliable End-to-End LLM Robot Programming through Expert-Informed Critics
End-user robot programming grants users the flexibility to re-task robots in situ, yet it remains challenging for novices due to the need for specialized robotics knowledge. Large Language Models (LLMs) hold the potential to lower the barrier to robot programming by enabling task specification through natural language. However, current LLM-based approaches generate opaque, "black-box" code that is difficult to verify or debug, creating tangible safety and reliability risks in physical systems. We present RoboCritics, an approach that augments LLM-based robot programming with expert-informed motion-level critics. These critics encode robotics expertise to analyze motion-level execution traces for issues such as joint...
Research Paper
Accepted to HRI 2026
RoboCritics: Enabling Reliable End-to-End LLM Robot Programming through Expert-Informed Critics
End-user robot programming grants users the flexibility to re-task robots in situ, yet it remains challenging for novices due to the need for specialized robotics knowledge. Large Language Models (LLMs) hold the potential to lower the barrier to robot programming by enabling task specification through natural language. However, current LLM-based approaches generate opaque, “black-box” code that is difficult to verify…
SkillOrchestra: Learning to Route Agents via Skill Transfer
Compound AI systems promise capabilities beyond those of individual models, yet their success depends critically on effective orchestration. Existing routing approaches face two limitations: (1) input-level routers make coarse query-level decisions that ignore evolving task requirements; (2) RL-trained orchestrators are expensive to adapt and often suffer from routing collapse, repeatedly invoking one strong but costly option in multi-turn scenarios. We introduce SkillOrchestra, a framework for skill-aware orchestration. Instead of directly learning a routing policy end-to-end, SkillOrchestra learns fine-grained skills from execution experience and models agent-specific competence and cost under those skills. At deployment, the orchestrator infers the skill demands of...
Research Paper
Accepted to NeurIPS 2026
SkillOrchestra: Learning to Route Agents via Skill Transfer
Compound AI systems promise capabilities beyond those of individual models, yet their success depends critically on effective orchestration. Existing routing approaches face two limitations: (1) input-level routers make coarse query-level decisions that ignore evolving task requirements; (2) RL-trained orchestrators are expensive to adapt and often suffer from routing collapse, repeatedly invoking one strong but costly option in multi-turn scenarios. We…
CARE: Confounder-Aware Aggregation for Reliable LLM Evaluation
LLM-as-a-judge ensembles are the standard paradigm for scalable evaluation, but their aggregation mechanisms suffer from a fundamental flaw: they implicitly assume that judges provide independent estimates of true quality. However, in practice, LLM judges exhibit correlated errors caused by shared latent confounders -- such as verbosity, stylistic preferences, or training artifacts -- causing standard aggregation rules like majority vote or averaging to provide little gain or even amplify systematic mistakes. To address this, we introduce CARE, a confounder-aware aggregation framework that explicitly models LLM judge scores as arising from both a latent true-quality signal and shared confounding factors. Rather than...
Research Paper
Accepted to ICML 2026
CARE: Confounder-Aware Aggregation for Reliable LLM Evaluation
LLM-as-a-judge ensembles are the standard paradigm for scalable evaluation, but their aggregation mechanisms suffer from a fundamental flaw: they implicitly assume that judges provide independent estimates of true quality. However, in practice, LLM judges exhibit correlated errors caused by shared latent confounders — such as verbosity, stylistic preferences, or training artifacts — causing standard aggregation rules like majority vote or…
