Abstract
Extending the context window and adapting to a new rotary positional embedding scaling parameter improve repository-level code completion in OpenCoder, achieving performance comparable to larger models with less data.
Repository-level pretraining is commonly used to enable large language models for code to leverage codebase-wide context. This enhances their ability to generate accurate and context-aware code completions. In this work, we investigate how different repository-processing strategies affect in-context learning in OpenCoder, a 1.5B-parameter model. We extend its context window from 4,096 to 16,384 tokens by training on additional 1B tokens of curated repository-level data. Despite relying on a smaller dataset than competing models (which often use hundreds of billions of tokens), our model achieves comparable performance on the Long Code Arena benchmark. We find that various repository-processing techniques yield similarly strong results, with the primary gain coming from adapting to a new rotary positional embedding (RoPE) scaling parameter. Finally, we show that a simpler file-level training approach at the original sequence length remains highly effective, opening up repository-level code completion research to settings with more constrained data and compute resources.
Community
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- Beyond More Context: How Granularity and Order Drive Code Completion Quality (2025)
- SynthCoder: A Synthetical Strategy to Tune LLMs for Code Completion (2025)
- CodeRAG: Finding Relevant and Necessary Knowledge for Retrieval-Augmented Repository-Level Code Completion (2025)
- Mellum: Production-Grade in-IDE Contextual Code Completion with Multi-File Project Understanding (2025)
- LongCodeZip: Compress Long Context for Code Language Models (2025)
- Repository-Aware File Path Retrieval via Fine-Tuned LLMs (2025)
- Challenge on Optimization of Context Collection for Code Completion (2025)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment:
@librarian-bot
recommend
Models citing this paper 32
Browse 32 models citing this paperDatasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper