README.md

---
license: cc-by-4.0
language:
- en
tags:
- streaming-cot
- chain-of-thought
- sft
- text
size_categories:
- 1K<n<10K
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train.parquet
  - split: eval
    path: data/eval.parquet
- config_name: high_quality
  data_files:
  - split: train
    path: data/high_quality_train.parquet
  - split: eval
    path: data/high_quality_eval.parquet
---

# LifeTextMultiTurnStreamingCoT

**Version:** vFinal

A multi-turn text dataset with streaming chain-of-thought reasoning for SFT. 9725 active training rows across daily-life, social, and productivity tasks.

## Schema

Top-level fields: `id, split, modality, turn_type, dialogue, input, streaming, target, taxonomy, quality, source, metadata`

- dialogue (list of turns with role/text)
- input (instruction, length_bucket)
- target (reasoning, answer, response)
- streaming (checkpoints with streaming_reasoning)
- taxonomy (category, subcategory, difficulty, intent_type)

### Target Format
```json
{
  "reasoning": "Natural language reasoning about the task/input...",
  "answer": "The actual task output...",
  "response": "Reasoning: ...\n\nAnswer: ..."
}
```

Use `target.response` for SFT training. It includes both reasoning and final answer.

## Quality

| Metric | Value |
|--------|-------|
| Active rows | 9,725 |
| Train | 7,740 |
| Eval | 1,985 |
| High quality | 9,619 |
| SFT-ready | 100.0% |
| Target grounded | 99.3% |

## High-Quality Configuration

The `high_quality` config contains a filtered subset of default rows where `quality.sft_ready = true` and `quality.is_high_quality = true`. It is not additional unique data.

## Limitations

- Text-only dataset.
- 
- Natural-language reasoning is template-generated, not LLM-written.
- Row counts reflect quality-filtered active splits suitable for direct SFT usage.

## Usage

```python
from datasets import load_dataset

# Load default config
ds = load_dataset("skyzhou06/LifeTextMultiTurnStreamingCoT")

# Load high-quality subset
ds_hq = load_dataset("skyzhou06/LifeTextMultiTurnStreamingCoT", "high_quality")
```