About

Christopher Re is an Associate Professor of Computer Science at Stanford University and is affiliated with the Center for Artificial Intelligence in Medicine & Imaging (AIMI). His work focuses on the application of artificial intelligence to medicine and imaging, contributing to the advancement of healthcare through innovative AI research. As a faculty member at Stanford, he is involved in various educational and research initiatives aimed at integrating AI technologies into medical practice and imaging sciences.

Research topics

• Computer science
• Artificial intelligence
• Machine learning
• Algorithm
• Theoretical computer science

Selected publications

• Development of a Ground Test Protocol for Human-in-the-Loop Evaluations of Drivable Lunar Terrain Vehicle Engineering Development Units

  2026-03-07

  article

  The National Aerospace and Astronautical Administration's (NASA) Lunar Terrain Vehicle (LTV) seeks commercial contractors to develop the next generation unpressurized rover to support two crewmembers and equipment for lunar surface expeditions. NASA research engineers, in collaboration with subject matter experts, utilized NASA Johnson Space Center's Ground Test Unit (GTU) vehicle to develop a set of ground-based driving tasks and build an evaluation protocol for assessment of crewed LTV operations. The purpose of the protocol is to provide a controlled, test track-like environment to methodically evaluate the crewed capabilities of an LTV and provide recommendations for design improvements. Areas evaluated through human-in-the-loop (HITL) testing included the vehicle operations, operator workload, human machine interfaces, and seated comfort, among others. After reviewing prior HITL vehicle testing from the military, automotive, aerospace industries, and NASA, a task battery was designed that encompassed all expected driving operations and environments that the LTV will need to operate in. This included driving on flat sections, sloped surfaces, overcoming obstacles such as rocks and craters, and navigating through challenging terrain that requires precise driving. These terrain types and scenarios were demonstrated in the JSC Rock Yard, an outdoor facility which has simulated craters, a hill, and boulder field. While future testing will utilize pressurized suits to evaluate the drivable vehicles, this paper will summarize the methods, metrics, and lessons learned from initial shirtsleeve testing using the GTU.

  PublisherDOI

• SMOOTHIE: Label Free Language Model Routing

  Qeios · 2025-01-09

  preprintOpen accessSenior author

  Large language models (LLMs) are increasingly used in applications where LLM inputs may span many different tasks. Recent work has found that the choice of LLM is consequential, and different LLMs may be good for different input samples. Prior approaches have thus explored how engineers might select an LLM to use for each sample (i.e. _routing_). While existing routing methods mostly require training auxiliary models on human-annotated data, our work explores whether it is possible to perform _unsupervised_ routing. We propose SMOOTHIE, a weak supervision-inspired routing approach that requires no labeled data. Given a set of outputs from different LLMs, SMOOTHIE constructs a latent variable graphical model over embedding representations of observable LLM outputs and unknown “true” outputs. Using this graphical model, we estimate sample-dependent quality scores for each LLM, and route each sample to the LLM with the highest corresponding score. We find that SMOOTHIE’s LLM quality-scores correlate with ground-truth model quality (correctly identifying the optimal model on 9/14 tasks), and that SMOOTHIE outperforms baselines for routing by up to 10 points accuracy.

  PublisherOA PDFDOI

• Systems and Algorithms for Convolutional Multi-Hybrid Language Models at Scale

  ArXiv.org · 2025-02-25 · 3 citations

  articleOpen access

  We introduce convolutional multi-hybrid architectures, with a design grounded on two simple observations. First, operators in hybrid models can be tailored to token manipulation tasks such as in-context recall, multi-token recall, and compression, with input-dependent convolutions and attention offering complementary performance. Second, co-designing convolution operators and hardware-aware algorithms enables efficiency gains in regimes where previous alternative architectures struggle to surpass Transformers. At the 40 billion parameter scale, we train end-to-end 1.2 to 2.9 times faster than optimized Transformers, and 1.1 to 1.4 times faster than previous generation hybrids. On H100 GPUs and model width 4096, individual operators in the proposed multi-hybrid StripedHyena 2 architecture achieve two-fold throughput improvement over linear attention and state-space models. Multi-hybrids excel at sequence modeling over byte-tokenized data, as demonstrated by the Evo 2 line of models. We discuss the foundations that enable these results, including architecture design, overlap-add blocked kernels for tensor cores, and dedicated all-to-all and point-to-point context parallelism strategies.

  PublisherOA PDF

• Cartridges: Lightweight and general-purpose long context representations via self-study

  ArXiv.org · 2025-06-06

  preprintOpen accessSenior author

  Large language models are often used to answer queries grounded in large text corpora (e.g. codebases, legal documents, or chat histories) by placing the entire corpus in the context window and leveraging in-context learning (ICL). Although current models support contexts of 100K-1M tokens, this setup is costly to serve because the memory consumption of the KV cache scales with input length. We explore an alternative: training a smaller KV cache offline on each corpus. At inference time, we load this trained KV cache, which we call a Cartridge, and decode a response. Critically, the cost of training a Cartridge can be amortized across all the queries referencing the same corpus. However, we find that the naive approach of training the Cartridge with next-token prediction on the corpus is not competitive with ICL. Instead, we propose self-study, a training recipe in which we generate synthetic conversations about the corpus and train the Cartridge with a context-distillation objective. We find that Cartridges trained with self-study replicate the functionality of ICL, while being significantly cheaper to serve. On challenging long-context benchmarks, Cartridges trained with self-study match ICL performance while using 38.6x less memory and enabling 26.4x higher throughput. Self-study also extends the model's effective context length (e.g. from 128k to 484k tokens on MTOB) and surprisingly, leads to Cartridges that can be composed at inference time without retraining.

  PublisherOA PDFDOI

• Minions: Cost-efficient Collaboration Between On-device and Cloud Language Models

  ArXiv.org · 2025-02-21

  preprintOpen accessSenior author

  We investigate an emerging setup in which a small, on-device language model (LM) with access to local data communicates with a frontier, cloud-hosted LM to solve real-world tasks involving financial, medical, and scientific reasoning over long documents. Can a local-remote collaboration reduce cloud inference costs while preserving quality? First, we consider a naive collaboration protocol where the local and remote models simply chat back and forth. Because only the local model reads the full context, this protocol achieves a 30.4x reduction in remote costs, but recovers only 87% of the performance of the frontier model. We identify two key limitations of this protocol: the local model struggles to (1) follow the remote model's multi-step instructions and (2) reason over long contexts. Motivated by these observations, we study an extension of this protocol, coined MinionS, in which the remote model decomposes the task into easier subtasks over shorter chunks of the document, that are executed locally in parallel. MinionS reduces costs by 5.7x on average while recovering 97.9% of the performance of the remote model alone. Our analysis reveals several key design choices that influence the trade-off between cost and performance in local-remote systems.

  PublisherOA PDFDOI

• HMAR: Efficient Hierarchical Masked Auto-Regressive Image Generation

  ArXiv.org · 2025-06-04

  preprintOpen access

  Visual Auto-Regressive modeling (VAR) has shown promise in bridging the speed and quality gap between autoregressive image models and diffusion models. VAR reformulates autoregressive modeling by decomposing an image into successive resolution scales. During inference, an image is generated by predicting all the tokens in the next (higher-resolution) scale, conditioned on all tokens in all previous (lower-resolution) scales. However, this formulation suffers from reduced image quality due to the parallel generation of all tokens in a resolution scale; has sequence lengths scaling superlinearly in image resolution; and requires retraining to change the sampling schedule. We introduce Hierarchical Masked Auto-Regressive modeling (HMAR), a new image generation algorithm that alleviates these issues using next-scale prediction and masked prediction to generate high-quality images with fast sampling. HMAR reformulates next-scale prediction as a Markovian process, wherein the prediction of each resolution scale is conditioned only on tokens in its immediate predecessor instead of the tokens in all predecessor resolutions. When predicting a resolution scale, HMAR uses a controllable multi-step masked generation procedure to generate a subset of the tokens in each step. On ImageNet 256x256 and 512x512 benchmarks, HMAR models match or outperform parameter-matched VAR, diffusion, and autoregressive baselines. We develop efficient IO-aware block-sparse attention kernels that allow HMAR to achieve faster training and inference times over VAR by over 2.5x and 1.75x respectively, as well as over 3x lower inference memory footprint. Finally, HMAR yields additional flexibility over VAR; its sampling schedule can be changed without further training, and it can be applied to image editing tasks in a zero-shot manner.

  PublisherOA PDFDOI

• KernelBench: Can LLMs Write Efficient GPU Kernels?

  ArXiv.org · 2025-02-14

  preprintOpen access

  Efficient GPU kernels are crucial for building performant machine learning architectures, but writing them is a time-consuming challenge that requires significant expertise; therefore, we explore using language models (LMs) to automate kernel generation. We introduce KernelBench, an open-source framework for evaluating LMs' ability to write fast and correct kernels on a suite of 250 carefully selected PyTorch ML workloads. KernelBench represents a real-world engineering environment and making progress on the introduced benchmark directly translates to faster practical kernels. We introduce a new evaluation metric fast_p, which measures the percentage of generated kernels that are functionally correct and offer a speedup greater than an adjustable threshold p over baseline. Our experiments across various state-of-the-art models and test-time methods show that frontier reasoning models perform the best out of the box but still fall short overall, matching the PyTorch baseline in less than 20% of the cases. While we show that results can improve by leveraging execution and profiling feedback during iterative refinement, KernelBench remains a challenging benchmark, with its difficulty increasing as we raise speedup threshold p.

  PublisherOA PDFDOI

• Intelligence per Watt: Measuring Intelligence Efficiency of Local AI

  ArXiv.org · 2025-11-11

  preprintOpen accessSenior author

  Large language model (LLM) queries are predominantly processed by frontier models in centralized cloud infrastructure. Demand growth strains this paradigm faster than providers can scale. Two advances create an opportunity to rethink it: small, local LMs (<=20B active parameters) now achieve competitive performance to frontier models on many tasks, and local accelerators (e.g., Apple M4 Max) can host these models at interactive latencies. This raises the question: can local inference viably redistribute demand from centralized infrastructure? This requires measuring both whether local LMs can accurately answer real-world queries and whether they can do so efficiently on power-constrained devices (e.g., laptops). We propose intelligence per watt (IPW), task accuracy per unit of power, as a unified metric for the capability and efficiency of local inference across model-accelerator configurations. We evaluate 20+ state-of-the-art local LMs, 8 hardware accelerators (local and cloud), and 1M real-world single-turn chat and reasoning queries. For each query, we measure accuracy (local LM win rate against frontier models), energy, latency, and power. We find three key results. First, local LMs successfully answer 88.7% of these queries, with accuracy varying by domain. Second, longitudinal analysis from 2023-2025 shows IPW improved 5.3x, driven by both algorithmic and accelerator advances, with locally-serviceable query coverage rising from 23.2% to 71.3%. Third, local accelerators achieve at least 1.4x lower IPW than cloud accelerators running identical models, revealing significant headroom for local accelerator optimization. These findings demonstrate that local inference can meaningfully redistribute demand from centralized infrastructure for a substantial subset of queries, with IPW serving as the critical metric for tracking this transition.

  PublisherOA PDFDOI

• CodeMonkeys: Scaling Test-Time Compute for Software Engineering

  ArXiv.org · 2025-01-24

  preprintOpen access

  Scaling test-time compute is a promising axis for improving LLM capabilities. However, test-time compute can be scaled in a variety of ways, and effectively combining different approaches remains an active area of research. Here, we explore this problem in the context of solving real-world GitHub issues from the SWE-bench dataset. Our system, named CodeMonkeys, allows models to iteratively edit a codebase by jointly generating and running a testing script alongside their draft edit. We sample many of these multi-turn trajectories for every issue to generate a collection of candidate edits. This approach lets us scale "serial" test-time compute by increasing the number of iterations per trajectory and "parallel" test-time compute by increasing the number of trajectories per problem. With parallel scaling, we can amortize up-front costs across multiple downstream samples, allowing us to identify relevant codebase context using the simple method of letting an LLM read every file. In order to select between candidate edits, we combine voting using model-generated tests with a final multi-turn trajectory dedicated to selection. Overall, CodeMonkeys resolves 57.4% of issues from SWE-bench Verified using a budget of approximately 2300 USD. Our selection method can also be used to combine candidates from different sources. Selecting over an ensemble of edits from existing top SWE-bench Verified submissions obtains a score of 66.2% and outperforms the best member of the ensemble on its own. We fully release our code and data at https://scalingintelligence.stanford.edu/pubs/codemonkeys.

  PublisherOA PDFDOI

• HMAR: Efficient Hierarchical Masked Auto-Regressive Image Generation

  2025-06-10 · 1 citations

  article

  Visual Auto-Regressive modeling (VAR) has shown promise in bridging the speed and quality gap between autoregressive image models and diffusion models. VAR reformulates autoregressive modeling by decomposing an image into successive resolution scales. During inference, an image is generated by predicting all the tokens in the next (higher-resolution) scale, conditioned on all tokens in all previous (lower-resolution) scales. However, this formulation suffers from reduced image quality due to the parallel generation of all tokens in a resolution scale; has sequence lengths scaling superlinearly in image resolution; and requires retraining to change the sampling schedule.We introduce Hierarchical Masked AutoRegressive modeling (HMAR), a new image generation algorithm that alleviates these issues using next-scale prediction and masked prediction to generate high-quality images with fast sampling. HMAR reformulates next-scale prediction as a Markovian process, wherein the prediction of each resolution scale is conditioned only on tokens in its immediate predecessor instead of the tokens in all predecessor resolutions. When predicting a resolution scale, HMAR uses a controllable multi-step masked generation procedure to generate a subset of the tokens in each step. On ImageNet 256 × 256 and 512 × 512 benchmarks, HMAR models match or outperform parameter-matched VAR, diffusion, and autoregressive baselines. We develop efficient IO-aware block-sparse attention kernels that allow HMAR to achieve faster training and inference times over VAR by over 2.5× and 1.75× respectively, as well as over 3× lower inference memory foot-print. Finally, HMAR yields additional flexibility over VAR; its sampling schedule can be changed without further training, and it can be applied to image editing tasks in a zero-shot manner.

  PublisherDOI

Frequent coauthors

• Atri Rudra

  44 shared

• Dan Suciu

  University of Washington

  43 shared

• Christopher De

  40 shared

• Karan Goel

  Stanford University

  34 shared

• Tri Dao

  30 shared

• Kunle Olukotun

  Stanford University

  27 shared

• Sen Wu

  State Key Laboratory of Solidification Processing

  27 shared

• Jesse Vig

  22 shared