Publications (18)
What Makes Large Language Models Reason in (Multi-Turn) Code Generation?
Kunhao Zheng, Juliette Decugis, Jonas Gehring +3
Prompting techniques such as chain-of-thought have established themselves as a popular vehicle for improving the outputs of large language models (LLMs). For code generation, howev…
RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learning
Jonas Gehring, Kunhao Zheng, Jade Copet +4
Large language models (LLMs) deployed as agents solve user-specified tasks over multiple steps while keeping the required manual engagement to a minimum. Crucially, such LLMs need…
Leveraging Demonstrations with Latent Space Priors
Jonas Gehring, Deepak Gopinath, Jungdam Won +3
Demonstrations provide insight into relevant state or action space regions, bearing great potential to boost the efficiency and practicality of reinforcement learning agents. In th…
STARDATA: A StarCraft AI Research Dataset
Zeming Lin, Jonas Gehring, Vasil Khalidov +1
We release a dataset of 65646 StarCraft replays that contains 1535 million frames and 496 million player actions. We provide full game state data along with the original replays th…
Towards a Neural Debugger for Python
Maximilian Beck, Jonas Gehring, Jannik Kossen +1
Training large language models (LLMs) on Python execution traces grounds them in code execution and enables the line-by-line execution prediction of whole Python programs, effectiv…
High-Level Strategy Selection under Partial Observability in StarCraft: Brood War
Jonas Gehring, Da Ju, Vegard Mella +3
We consider the problem of high-level strategy selection in the adversarial setting of real-time strategy games from a reinforcement learning perspective, where taking an action co…
Large Language Models for Compiler Optimization
Chris Cummins, Volker Seeker, Dejan Grubisic +8
We explore the novel application of Large Language Models to code optimization. We present a 7B-parameter transformer model trained from scratch to optimize LLVM assembly for code…
The Larger the Better? Improved LLM Code-Generation via Budget Reallocation
Michael Hassid, Tal Remez, Jonas Gehring +2
It is a common belief that large language models (LLMs) are better than smaller-sized ones. However, larger models also require significantly more time and compute during inference…
CWM: An Open-Weights LLM for Research on Code Generation with World Models
FAIR CodeGen team, Jade Copet, Quentin Carbonneaux +48
We release Code World Model (CWM), a 32-billion-parameter open-weights LLM, to advance research on code generation with world models. To improve code understanding beyond what can…
Toward Training Superintelligent Software Agents through Self-Play SWE-RL
Yuxiang Wei, Zhiqing Sun, Emily McMilin +6
While current software agents powered by large language models (LLMs) and agentic reinforcement learning (RL) can boost programmer productivity, their training data (e.g., GitHub i…
Meta Large Language Model Compiler: Foundation Models of Compiler Optimization
Chris Cummins, Volker Seeker, Dejan Grubisic +4
Large Language Models (LLMs) have demonstrated remarkable capabilities across a variety of software engineering and coding tasks. However, their application in the domain of code a…
Hierarchical Skills for Efficient Exploration
Jonas Gehring, Gabriel Synnaeve, Andreas Krause +1
In reinforcement learning, pre-trained low-level skills have the potential to greatly facilitate exploration. However, prior knowledge of the downstream task is required to strike…
Code Llama: Open Foundation Models for Code
Baptiste Rozière, Jonas Gehring, Fabian Gloeckle +23
We release Code Llama, a family of large language models for code based on Llama 2 providing state-of-the-art performance among open models, infilling capabilities, support for lar…
Convolutional Sequence to Sequence Learning
Jonas Gehring, Michael Auli, David Grangier +2
The prevalent approach to sequence to sequence learning maps an input sequence to a variable length output sequence via recurrent neural networks. We introduce an architecture base…
The KoLMogorov Test: Compression by Code Generation
Ori Yoran, Kunhao Zheng, Fabian Gloeckle +3
Compression is at the heart of intelligence. A theoretically optimal way to compress any sequence of data is to find the shortest program that outputs that sequence and then halts.…
Forward Modeling for Partial Observation Strategy Games - A StarCraft Defogger
Gabriel Synnaeve, Zeming Lin, Jonas Gehring +5
We formulate the problem of defogging as state estimation and future state prediction from previous, partial observations in the context of real-time strategy games. We propose to…
A Convolutional Encoder Model for Neural Machine Translation
Jonas Gehring, Michael Auli, David Grangier +1
The prevalent approach to neural machine translation relies on bi-directional LSTMs to encode the source sentence. In this paper we present a faster and simpler architecture based…
Extrapolative Weight Averaging Reveals Correctness-Efficiency Frontiers in Code RL
Kunhao Zheng, Pierre Chambon, Juliette Decugis +4
Linear interpolation between fine-tuned checkpoints has been shown to trace the Pareto front between competing objectives, but whether extrapolative weight averaging can extend suc…