Sign in to view source links and access this dataset
Description
692,587 rows of Japanese-Chinese chain-of-thought reasoning data derived from a light novel source stream, with reasoning_content preserved. The dataset was created by OpenSakura and includes splits for training, validation, test, arena, and reserve evaluation. It was last updated on Hugging Face in May 2026.
Use Cases
Training language models on step-by-step reasoning based on the preserved reasoning_content.
Evaluating model performance on Japanese-Chinese reasoning tasks using the provided arena and test splits.
Fine-tuning models for instruction-following or problem-solving in a bilingual context.
Benchmarking the quality of generated chain-of-thought outputs against a structured dataset.
Strengths
692,587 rows provide a substantial corpus for training.
The dataset includes dedicated splits for training (162 files), validation (24 files), test (23 files), and arena/reserve evaluation (24 files), supporting robust model development.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
The description metadata is limited; actual data quality and content require manual inspection after download.
Provenance
Source
OpenSakura
Collection Method
Derived from a light novel source stream, with reasoning_content preserved.
Freshness
Last updated 2026-05-22 15:18:00; freshness should be verified.
License is unknown; users should verify permissions before use.