march228/grok-reflection-cot-ru

---
language:
- ru
license: other
pretty_name: Grok Reflection COT RU
tags:
- russian
- synthetic
- reasoning
- chain-of-thought
- dialogue
task_categories:
- text-generation
size_categories:
- 1K<n<10K
---

# march228/grok-reflection-cot-ru

Russian synthetic dataset with question, internal thought text, and final answer.

## What is inside

- Rows: `4190`
- Split: `train`
- Main fields:
- `question`
- `thought_text`
- `answer`
- `thought1..thought5`
- `model`
- `task_type`
- `reflection_count`

## Format

The dataset is stored as `train.jsonl`.

`thought_text` is the joined internal monologue with blank lines between thought blocks.
`thought1..thought5` preserve the original segmented form from the SQLite source.

## Intended use

- reasoning-style experiments
- Russian CoT / inner-monologue finetuning
- analysis of reflection-like intermediate reasoning

## Caveats

- The dataset is synthetic / teacher-generated.
- It is not a factual benchmark.
- Provenance includes model-generated content; check legal and policy constraints before broad reuse.
- `license: other` is intentionally conservative until a final public license decision is made.

## Load

```python
from datasets import load_dataset

ds = load_dataset("march228/grok-reflection-cot-ru")
print(ds["train"][0])
```