---
title: openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16
canonical_url: "https://www.modelscope.cn/datasets/marin-community/openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16"
md_url: "https://www.modelscope.cn/datasets/marin-community/openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16.md"
repository: marin-community/openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16
last_updated: 2026-06-27
license: apache-2.0
storage_size: "61 GB"
downloads: 165
stars: 0
---

# openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16

> openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16 - marin-community 在 ModelScope 开源的数据集。OpenThoughts-4 Science SDG: Qwen3-30B-A3B-Thinking-2507 (n=8, top-16 logprobs)

marin-community/openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16 是 ModelScope 魔搭社区上的数据集，存储大小 61 GB，采用 apache-2.0 许可。

- **Repository**: marin-community/openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16
- **License**: apache-2.0
- **Storage size**: 61 GB
- **Downloads**: 165
- **Stars**: 0
- **Last updated**: 2026-06-27

Source: https://www.modelscope.cn/datasets/marin-community/openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16

---

# OpenThoughts-4 Science SDG: Qwen3-30B-A3B-Thinking-2507 (n=8, top-16 logprobs)

Synthetic generations from
[**Qwen/Qwen3-30B-A3B-Thinking-2507**](https://huggingface.co/Qwen/Qwen3-30B-A3B-Thinking-2507)
on the [Marin OpenThoughts-4 science SDG prompt
set](https://huggingface.co/datasets/marin-community/hero-run-4-science-sdg-prompts-boxed-n8).
Each prompt is sampled `n=8` times, and for every generated token the dataset
stores the chosen-token log probability plus the **top-16** log probabilities
over the vocabulary, enabling distillation, KL-style fine-tuning,
reranking, and uncertainty analysis.

## Generation setup

| Field | Value |
|---|---|
| Generator model | `Qwen/Qwen3-30B-A3B-Thinking-2507` |
| Source prompts | `marin-community/hero-run-4-science-sdg-prompts-boxed-n8` (26,041 unique prompts) |
| Samples per prompt (`n`) | 8 |
| Logprobs returned (`k`) | 16 (top-k vocab logprobs per generated token) |
| Max generated tokens | 32,768 |
| Max model length | 34,816 |
| Temperature | 0.8 |
| Inference engine | vLLM on TPU v6e-4, `tensor_parallel_size=4` |
| Producer | [marin-community/marin](https://github.com/marin-community/marin) — `experiments/sdg/science/qwen3-30b-a3b-thinking-2507/sdg_ot4_30k_science_qwen3_30b_a3b_thinking_32768_tokens.py` |

## Schema

The dataset is a flattened parquet table with **one row per
`(prompt, sample_index)`** pair. With 26,041 unique prompts and `n=8` samples
each, the dataset contains **208,328 rows** total.

### Identifier columns

| Column | Type | Description |
|---|---|---|
| `prompt_index` | int64 | 0-based index of the source prompt within the prompt set (0 … 26,040) |
| `response_index` | int64 | 0-based sample index within that prompt (0 … 7) |
| `_unique_row_id` | string | Stable globally-unique id for the `(prompt, response)` pair, copied from the source prompt set; safe key for joins and dedup |
| `instruction_seed` | string | The original natural-language science problem from OpenThoughts-4 (before any chat templating) |
| `generation_prompt` | string | The chat-templated prompt actually fed to vLLM (Qwen3 chat template applied to `instruction_seed`) |

A given prompt is fully identified by either `prompt_index` or `_unique_row_id`;
the 8 samples for a prompt share both keys and differ only in `response_index`.

### Generation columns

| Column | Type | Description |
|---|---|---|
| `generated_text` | string | Decoded model response (everything after the chat template's assistant turn) |
| `generated_token_ids` | list[int32], length `T` | Token ids of the generated response |
| `generated_token_logprobs` | list[float32], length `T` | Log probability of each chosen token under the model |
| `generated_top_logprob_token_ids` | list[int32], length `T × k` | **Flattened** top-k candidate token ids at each step |
| `generated_top_logprobs` | list[float32], length `T × k` | **Flattened** log probabilities of those top-k candidates at each step |

Where `T` = number of generated tokens for that row and `k = 16`.

### Reshaping the flattened top-k arrays

vLLM was run with `flat_logprobs=True` for compactness; the per-step
top-k arrays are stored as 1-D lists. To recover the standard
`(T, k)` shape:

```python
import numpy as np
import pyarrow.parquet as pq

df = pq.read_table("part-000000.parquet").to_pandas()
row = df.iloc[0]
T = len(row["generated_token_ids"])
k = 16
top_ids   = np.asarray(row["generated_top_logprob_token_ids"]).reshape(T, k)
top_logps = np.asarray(row["generated_top_logprobs"]).reshape(T, k)
# top_ids[t, :]   = top-16 candidate token ids at step t
# top_logps[t, :] = log p of those candidates at step t (sorted high → low by vLLM)
# the chosen token may or may not appear in the top-k; use generated_token_logprobs
# for the chosen-token logprob.
```

### File layout

The dataset is sharded into ~3,600 parquet files named `part-NNNNNN.parquet`.
Use the `default` config to load the full `train` split.

## Companion datasets

| Slice | Generator |
|---|---|
| science | Qwen3-30B-A3B-Thinking-2507 (this dataset) |
| science | [Qwen3-32B](https://huggingface.co/datasets/marin-community/openthoughts4-science-26041-prompts-qwen3-32b-n8-flattened-logprobs-k16) |
| science | [Qwen3-4B](https://huggingface.co/datasets/marin-community/openthoughts4-science-26041-prompts-qwen3-4b-n8-flattened-logprobs-k16) |
| science | Gemma-4-31B-IT *(forthcoming)* |
| code | [Qwen3-30B-A3B-Thinking-2507](https://huggingface.co/datasets/marin-community/openthoughts4-code-9168-prompts-qwen3-30b-a3B-thinking-2507-n16-flattened-logprobs-k16) / [Qwen3-32B](https://huggingface.co/datasets/marin-community/openthoughts4-code-9168-prompts-qwen3-32b-n16-flattened-logprobs-k16) / [Qwen3-4B](https://huggingface.co/datasets/marin-community/openthoughts4-code-9168-prompts-qwen3-4b-n16-flattened-logprobs-k16) |

## License

Released under Apache 2.0. The underlying generator model
(`Qwen/Qwen3-30B-A3B-Thinking-2507`) is governed by its own license; consult
the model card before redistribution.

## Citation

If you use this dataset, please cite the
[Marin project](https://github.com/marin-community/marin) and the
[Qwen3 technical report](https://arxiv.org/abs/2505.09388).