Dataset · nvidia
SFT corpus for terminal agents
nvidia/Nemotron-Terminal-Corpus
English SFT dataset from NVIDIA with trajectories for training LLMs to work in the terminal.
What's inside
The dataset contains text trajectories for supervised fine-tuning of terminal agents. It was collected by NVIDIA using the Terminal-Task-Gen pipeline, which combines dataset adaptation and synthetic task generation across various domains. English language, size from 100k to 1M examples.
What you can do with it
- SFT LLM for executing terminal commands
- LoRA fine-tuning of a terminal agent
- Evaluating agent models on terminal tasks
- Tasks
- Question answering
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 11K
- Likes
- 155
Models trained on it
- open-athena/Qwen3.5-35B-A3B-Chat-Aligned-Base-2026.10.31Text generation 1.1K
- open-athena/Qwen3-30B-A3B-Chat-Aligned-Base-2026.10.30Text generation 938
- open-athena/Nemotron-3-Nano-30B-A3B-Chat-Aligned-Base-2026.10.30Text generation 723
- open-athena/Snowball-67B-A2B-Chat-Aligned-Base-2026.10.22Text generation 338
- mradermacher/LFM2.5-2.6B-Terminal-SFT-i1-GGUF 287
- Luminia/MiniCPM5-1B-Agent-GGUFText generation 267
- mradermacher/LFM2.5-2.6B-Terminal-SFT-GGUF 103
- shr3y/Qwen3-0.6B-Agentic-Terminal-GGUF 102
How to load it
from datasets import load_dataset
ds = load_dataset("nvidia/Nemotron-Terminal-Corpus")