[Announcement] 1,841 samples - Physics Science dialogue dataset now available
Physics Science - 1,841 High-Quality Chinese Dialogue Samples
What is this dataset?
中文物理科学深度对话，涵盖量子引力、万物理论、相对论、量子力学前沿讨论.
All conversations are original, human-authored deep dialogues — not synthetic GPT-generated data. The conversations span multiple years of in-depth discussion across 14 academic and creative disciplines.
Key Features
- Language: Chinese (zh)
- Samples: 1,841 dialogue pairs in instruction-output format
- Format: JSONL (ready for SFT training)
- Quality: Real multi-turn deep reasoning, not template-based generation
- License: CC-BY-NC-SA-4.0 (non-commercial use free, commercial license available)
Why this dataset is different
Most Chinese dialogue datasets on HuggingFace are either:
- Machine-translated from English (losing nuance)
- Synthetic GPT outputs (lacking genuine reasoning depth)
- Short Q&A pairs (lacking multi-turn complexity)
This dataset contains genuine deep-dive conversations with real intellectual curiosity — questions that push boundaries, answers that cite multiple disciplines, and follow-up discussions that build on previous points.
Sample Preview
A 200-sample preview is available in the sample/ directory. Download the full dataset to access all 1,841 pairs.
Commercial Use
For commercial licensing, contact: yuexiangruiyue@gmail.com
Related Datasets
This dataset is part of a 14-dataset collection with 49,522 total samples across philosophy, psychology, medical, AI safety, physics, theology, art, and more.
Full collection: https://huggingface.co/collections/AngelWarmSmile123/deep-chinese-dialogue-datasets-6a4dda4b5d083f6eefb9c333
If you find this dataset useful, please click the Like button to help others discover it!