File size: 2,197 Bytes
7b03841
26ef827
7b03841
 
 
 
26ef827
7b03841
26ef827
 
 
7b03841
 
 
 
 
 
 
 
 
 
 
 
 
 
 
26ef827
7b03841
26ef827
7b03841
26ef827
7b03841
26ef827
7b03841
26ef827
7b03841
26ef827
 
 
 
 
 
7b03841
26ef827
7b03841
 
26ef827
 
7b03841
 
 
 
26ef827
7b03841
26ef827
7b03841
26ef827
 
 
 
 
 
 
 
7b03841
26ef827
7b03841
26ef827
7b03841
26ef827
7b03841
26ef827
 
 
 
7b03841
26ef827
7b03841
 
 
 
26ef827
7b03841
 
26ef827
7b03841
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
---
license: cc-by-4.0
language:
- en
tags:
- streaming-cot
- chain-of-thought
- sft
- audio
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
---

# LifeAudioMultiTurnStreamingCoT

**Version:** vFinal

A multi-turn audio dataset with streaming chain-of-thought reasoning for SFT. 13349 active training rows with audio references.

## Schema

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

- dialogue (list of turns with role/text)
- audio (path, reference, duration_sec, loadability)
- 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 | 13,349 |
| Train | 10,679 |
| Eval | 2,670 |
| High quality | 13,349 |
| SFT-ready | 100.0% |
| Target grounded | 100.0% |

## 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

- Audio is external reference only (HF URLs). No bundled audio bytes.
- 
- 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/LifeAudioMultiTurnStreamingCoT")

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