Nemotron-SWE-v1

← All datasets

Author: Distillio · License: cc-by-4.0 (a few subsets are bsd-3-clause — see HF dataset viewer) · Format: jsonl

0 downloads · 0 stars · 0 mirrors · size unknown · 1 file

59k agent trajectories targeting SWE-Bench-style repo navigation and issue-fixing.

Tags: tool-calling · agentic · code · swe-bench · open-harness · openhands

Description

59k agent trajectories collected using the OpenHands framework, targeting SWE-Bench-style repo navigation and issue-fixing. Issue statements sourced from SWE-Gym and R2E-Gym-Subset; trajectories synthesized with Qwen3-Coder-480B-A35B-Instruct. Explicitly ready for commercial use.

Download

Direct download: https://huggingface.co/datasets/nvidia/Nemotron-SWE-v1/resolve/main/data/r2e_gym.jsonl

Or via the API: curl -L -o file "https://huggingface.co/datasets/nvidia/Nemotron-SWE-v1/resolve/main/data/r2e_gym.jsonl"

Added 2026-08-11. Platform: Distillio — the open source training data collective.