---
license: other
pretty_name: Qwen3-4B PerfectBlend DeepSpec Rollout
task_categories:
  - text-generation
language:
  - multilingual
tags:
  - synthetic-data
  - knowledge-distillation
  - qwen3
  - deepspec
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/*.jsonl
      - split: eval
        path: eval/*.jsonl
---

# Qwen3-4B PerfectBlend DeepSpec Rollout

This dataset contains the complete DeepSpec-aligned Qwen3-4B
self-distillation rollout over the filtered PerfectBlend corpus. The seeded
95/5 split is published as separate `train` and `eval` splits.

## Splits

| Split | Conversations | Shards | Path |
|---|---:|---:|---|
| train | 1,349,860 | 128 | `data/*.jsonl` |
| eval | 71,046 | 64 | `eval/*.jsonl` |
| total | 1,420,906 | 192 | |

## Data construction

- Canonical filtered corpus: 1,420,906 conversations.
- Split: seed 42, 95% train / 5% eval.
- The original assistant messages are discarded. Every assistant turn is
  regenerated in order, and each generated answer is included in the context
  for subsequent turns in the same conversation.
- Generator: `Qwen/Qwen3-4B`, with thinking disabled, for both splits.
- Sampling: temperature 0.7, top-p 0.8, top-k 20, min-p 0.
- Maximum generated tokens per assistant turn: 4,096.
- No deduplication is applied.
- Canonical source content SHA256:
  `9d1376be67ae4c4708145d6939c4d8d49e467e5c4b78773fce633b81ea256613`.
- Reconstructed eval input SHA256:
  `291ba4429bf74e9dc3541cb9679a119bf924fa6edd4c9494d5f2bd523f97441d`.

## Schema

Each JSONL row is one conversation:

    {
      "id": 123,
      "conversations": [
        {"role": "user", "content": "..."},
        {"role": "assistant", "content": "..."}
      ],
      "status": "success"
    }

The `id` is the global ID from the canonical filtered dataset and is used for
exact-ID resume and coverage validation.

## Integrity

The published snapshot was validated before upload:

- Train: 128/128 shards, 1,349,860 rows, and 1,349,860 unique IDs.
- Eval: 64/64 shards, 71,046 rows, and 71,046 unique IDs.
- Eval generated IDs exactly equal the seeded source eval IDs.
- Train/eval ID intersection is zero; their union is exactly 1,420,906 IDs.
- No malformed JSON, failed statuses, duplicate IDs, missing IDs, or empty
  assistant responses were found.
- Train sorted global-ID SHA256:
  `625a89eada47d80063a25f31c920bca23272abfb86e6a433094b36a93be899be`.
- Remote file counts and byte sizes were checked after upload; generated file
  contents are retained in the validated ucsd-hopper snapshot.

Four records contain a whitespace-only user turn inherited unchanged from the
canonical source data.

Use of this generated dataset remains subject to the licenses and terms of the
upstream PerfectBlend data and Qwen3-4B model.

## Citation

This dataset is the training corpus behind **Beyond Parallel Blindness: Information Floors and
Model Gaps in Block Drafting** ([arXiv:2608.27339](https://arxiv.org/abs/2608.27339)), and behind
the drafters released at
[`TIE-Pilot/dspark-attnconv-block7-qwen3-4b`](https://huggingface.co/TIE-Pilot/dspark-attnconv-block7-qwen3-4b)
and
[`TIE-Pilot/deepspec-drafter-ablations`](https://huggingface.co/TIE-Pilot/deepspec-drafter-ablations).

```bibtex
@misc{qiang2026parallelblindnessinformationfloors,
      title={Beyond Parallel Blindness: Information Floors and Model Gaps in Block Drafting}, 
      author={Xinwei Qiang and Xiang Fang and Chang Chen and Zaifeng Pan and Yue Guan and Yufei Ding},
      year={2026},
      eprint={2608.27339},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2608.27339}, 
}
```