Marc-Alexandre Côté

Xingdi Yuan

2025-12-11

ArXiv (prépublication)

Gistify! Codebase-Level Understanding via Runtime Execution

Hyunji Lee

Minseon Kim

Chinmay Singh

Matheus Pereira

Atharv Sonwane

Isadora White

Elias Stengel-Eskin

Mohit Bansal

Zhengyan Shi

Xingdi Yuan

Lucas Caccia

As coding agents are increasingly deployed in large codebases, the need to automatically design challenging, codebase-level evaluation is ce… (voir plus)ntral. We propose Gistify, a task where a coding LLM must create a single, minimal, self-contained file that can reproduce a specific functionality of a codebase. The coding LLM is given full access to a codebase along with a specific entrypoint (e.g., a python command), and the generated file must replicate the output of the same command ran under the full codebase, while containing only the essential components necessary to execute the provided command. Success on Gistify requires both structural understanding of the codebase, accurate modeling of its execution flow as well as the ability to produce potentially large code patches. Our findings show that current state-of-the-art models struggle to reliably solve Gistify tasks, especially ones with long executions traces.

2025-10-29

ArXiv (prépublication)

BugPilot: Complex Bug Generation for Efficient Learning of SWE Skills

Atharv Sonwane

Isadora White

Hyunji Lee

Matheus Pereira

Lucas Caccia

Minseon Kim

Zhengyan Shi

Chinmay Singh

Marc-Alexandre Cot'e

Xingdi Yuan

2025-10-22

ArXiv (prépublication)

Learning to Solve Complex Problems via Dataset Decomposition

Wanru Zhao

Lucas Caccia

Zhengyan Shi

Minseon Kim

Xingdi Yuan

Weijia Xu

Curriculum learning is a class of training strategies that organizes the data being exposed to a model by difficulty, gradually from simpler… (voir plus) to more complex examples. This research explores a reverse curriculum generation approach that recursively decomposes complex datasets into simpler, more learnable components. We propose a teacher-student framework where the teacher is equipped with the ability to reason step-by-step, which is used to recursively generate easier versions of examples, enabling the student model to progressively master difficult tasks. We propose a novel scoring system to measure data difficulty based on its structural complexity and conceptual depth, allowing curriculum construction over decomposed data. Experiments on math datasets (MATH and AIME) demonstrate that models trained with curricula generated by our approach exhibit superior performance compared to standard training on original datasets.

2025-07-09

ICML.cc/2025/Workshop/AI4MATH (poster)

debug-gym: A Text-Based Environment for Interactive Debugging

Xingdi Yuan

Morgane M Moss

Charbel Feghali

Chinmay Singh

Darya Moldavskaya

Drew MacPhee

Lucas Caccia

Matheus Pereira

Minseon Kim

2025-03-27

ArXiv (prépublication)

debug-gym: A Text-Based Environment for Interactive Debugging

Xingdi Yuan

Morgane M Moss

Charbel Feghali

Chinmay Singh

Darya Moldavskaya

Drew MacPhee

Lucas Caccia

Matheus Pereira

Minseon Kim

2025-03-27

ArXiv (prépublication)

Sub-goal Distillation: A Method to Improve Small Language Agents

Maryam Hashemzadeh

Elias Stengel-Eskin

Sarath Chandar

While Large Language Models (LLMs) have demonstrated significant promise as agents in interactive tasks, their substantial computational req… (voir plus)uirements and restricted number of calls constrain their practical utility, especially in long-horizon interactive tasks such as decision-making or in scenarios involving continuous ongoing tasks. To address these constraints, we propose a method for transferring the performance of an LLM with billions of parameters to a much smaller language model (770M parameters). Our approach involves constructing a hierarchical agent comprising a planning module, which learns through Knowledge Distillation from an LLM to generate sub-goals, and an execution module, which learns to accomplish these sub-goals using elementary actions. In detail, we leverage an LLM to annotate an oracle path with a sequence of sub-goals towards completing a goal. Subsequently, we utilize this annotated data to fine-tune both the planning and execution modules. Importantly, neither module relies on real-time access to an LLM during inference, significantly reducing the overall cost associated with LLM interactions to a fixed cost. In ScienceWorld, a challenging and multi-task interactive text environment, our method surpasses standard imitation learning based solely on elementary actions by 16.7% (absolute). Our analysis highlights the efficiency of our approach compared to other LLM-based methods. Our code and annotated data for distillation can be found on GitHub.

2025-02-17

Proceedings of The 3rd Conference on Lifelong Learning Agents (publié)

Sub-goal Distillation: A Method to Improve Small Language Agents

Maryam Hashemzadeh

Elias Stengel-Eskin

Sarath Chandar

While Large Language Models (LLMs) have demonstrated significant promise as agents in interactive tasks, their substantial computational req… (voir plus)uirements and restricted number of calls constrain their practical utility, especially in long-horizon interactive tasks such as decision-making or in scenarios involving continuous ongoing tasks. To address these constraints, we propose a method for transferring the performance of an LLM with billions of parameters to a much smaller language model (770M parameters). Our approach involves constructing a hierarchical agent comprising a planning module, which learns through Knowledge Distillation from an LLM to generate sub-goals, and an execution module, which learns to accomplish these sub-goals using elementary actions. In detail, we leverage an LLM to annotate an oracle path with a sequence of sub-goals towards completing a goal. Subsequently, we utilize this annotated data to fine-tune both the planning and execution modules. Importantly, neither module relies on real-time access to an LLM during inference, significantly reducing the overall cost associated with LLM interactions to a fixed cost. In ScienceWorld, a challenging and multi-task interactive text environment, our method surpasses standard imitation learning based solely on elementary actions by 16.7% (absolute). Our analysis highlights the efficiency of our approach compared to other LLM-based methods. Our code and annotated data for distillation can be found on GitHub.

2024-05-04

ArXiv (prépublication)

Language-guided Skill Learning with Temporal Variational Inference

Haotian Fu

Pratyusha Sharma

Elias Stengel-Eskin

George Konidaris

Nicolas Le Roux

Xingdi Yuan

2024-05-01

ICML.cc/2024/Conference (poster)

Language-guided Skill Learning with Temporal Variational Inference

Haotian Fu

Pratyusha Sharma

Elias Stengel-Eskin

George Konidaris

Nicolas Le Roux

Xingdi Yuan

We present an algorithm for skill discovery from expert demonstrations. The algorithm first utilizes Large Language Models (LLMs) to propose… (voir plus) an initial segmentation of the trajectories. Following that, a hierarchical variational inference framework incorporates the LLM-generated segmentation information to discover reusable skills by merging trajectory segments. To further control the trade-off between compression and reusability, we introduce a novel auxiliary objective based on the Minimum Description Length principle that helps guide this skill discovery process. We test our system on BabyAI, a grid world navigation environment, as well as ALFRED, a household simulation environment.Our results demonstrate that agents equipped with our method can discover skills that help accelerate learning and outperform baseline skill learning approaches on new long-horizon tasks.

2024-03-11

ICLR.cc/2024/Workshop/LLMAgents (poster)

Joint Prompt Optimization of Stacked LLMs using Variational Inference

Eric Yuan

Xingdi Yuan

Matheus Pereira

Adam Trischler

Ziang Xiao

Arian Hosseini

Friederike Niedtner

Nicolas Le Roux

Large language models (LLMs) can be seen as atomic units of computation mapping sequences to a distribution over sequences. Thus, they can b… (voir plus)e seen as stochastic language layers in a language network, where the learnable parameters are the natural language prompts at each layer. By stacking two such layers and feeding the output of one layer to the next, we obtain a Deep Language Network (DLN). We first show how to effectively perform prompt optimization for a 1-Layer language network (DLN-1). Then, we present an extension that applies to 2-layer DLNs (DLN-2), where two prompts must be learned. The key idea is to consider the output of the first layer as a latent variable, which requires inference, and prompts to be learned as the parameters of the generative distribution. We first test the effectiveness of DLN-1 in multiple reasoning and natural language understanding tasks. Then, we show that DLN-2 can reach higher performance than a single layer, showing promise that we might reach comparable performance to GPT-4, even when each LLM in the network is smaller and less powerful.

From Words to Blocks: Building Objects by Grounding Language Models with Reinforcement Learning

Michael Ahn

Anthony Brohan

Noah Brown

liang Dai

Dan Su

Holy Lovenia Ziwei Bryan Wilie

Tiezheng Yu

Willy Chung

Quyet V. Do

Paul Barde

Tristan Karch

Derek Nowrouzezahrai

C. Bonial

Mitchell Abrams

David R. Traum

Hyung Won

Le Hou

Shayne Longpre

Yi Zoph

William Tay … (voir 32 de plus)

Eric Fedus

Xuezhi Li

Lasse Espeholt

Hubert Soyer

Remi Munos

Karen Si-801

Vlad Mnih

Tom Ward

Yotam Doron

Wenlong Huang

Pieter Abbeel

Deepak Pathak

Julia Kiseleva

Ziming Li

Mohammad Aliannejadi

Shrestha Mohanty

Maartje Ter Hoeve

Mikhail Burtsev

Alexey Skrynnik

Artem Zholus

A. Panov

Kavya Srinet

A. Szlam

Yuxuan Sun

Katja Hofmann