Paper 2602.21193
On Data Engineering for Scaling LLM Terminal Capabilities
- Published
- Feb 2026
- Research lab
- NVIDIA
- Citations
- 15
- GitHub
- Not linked
01 In brief
Summary
This paper addresses the lack of transparency in training data for terminal-capable LLMs by introducing Terminal-Task-Gen, a synthetic task generation pipeline, and Terminal-Corpus, a large-scale open-source dataset.
The authors fine-tune Qwen3 models (8B, 14B, 32B) to create Nemotron-Terminal, achieving substantial gains on Terminal-Bench 2.0: 8B improves from 2.5% to 13.0%, 14B from 4.0% to 20.2%, and 32B from 3.4% to 27.4%, matching or outperforming much larger models like Qwen3-Coder-480B.
The pipeline combines dataset adaptation (converting math, code, and SWE prompts) with synthetic task generation (seed-based and skill-based) using pre-built Docker images.
Experiments show that combining all data sources yields the best performance, no filtering of trajectories is optimal, and mixed training outperforms curriculum learning.
Scaling experiments confirm performance improves with more data.
The authors release model checkpoints and most synthetic datasets.
02 From the paper
Abstract
Despite rapid recent progress in the terminal capabilities of large language models, the training data strategies behind state-of-the-art terminal agents remain largely undisclosed. We address this gap through a systematic study of data engineering practices for terminal agents, making two key contributions: (1) Terminal-Task-Gen, a lightweight synthetic task generation pipeline that supports seed-based and skill-based task construction, and (2) a comprehensive analysis of data and training strategies, including filtering, curriculum learning, long context training, and scaling behavior. Our pipeline yields Terminal-Corpus, a large-scale open-source dataset for terminal tasks. Using this dataset, we train Nemotron-Terminal, a family of models initialized from Qwen3(8B, 14B, 32B) that achieve substantial gains on Terminal-Bench 2.0: Nemotron-Terminal-8B improves from 2.5% to 13.0% Nemotron-Terminal-14B improves from 4.0% to 20.2%, and Nemotron-Terminal-32B improves from 3.4% to 27.4%, matching the performance of significantly larger models. To accelerate research in this domain, we open-source our model checkpoints and most of our synthetic datasets at https://huggingface.co/collections/nvidia/nemotron-terminal.