---
pretty_name: "Magpie Qwen3 Bokm\xE5l"
language:
- nob
license: apache-2.0
license_name: Apache 2.0
tasks:
- general
- reasoning
- code
- instruction_following
creation_approach: synthetic
task_categories:
- conversational
- text-generation
---

# Dataset Card for Magpie Qwen3 Bokmål

<!-- START-SHORT DESCRIPTION -->
Synthetic Bokmål conversations generated by Qwen3-235B-A22B with Magpie.
<!-- END-SHORT DESCRIPTION -->

## Dataset Description

<!-- START-DESC-STATS -->
- **Number of samples**: 1.82K
- **Number of tokens (Llama 3)**: 2.34M
- **Average conversation length in tokens (min, max)**: 1.29K (360, 8.92K)
- **Average number of turns (min, max)**: 3.0 (3, 3)
<!-- END-DESC-STATS -->

This subset contains synthetic Norwegian prompts and responses for tasks such as
reasoning, coding, writing, and information seeking.

## Dataset Structure

<!-- START-SAMPLE -->
```py
{
  "id": "magpie-qwen3-bokmaal_00000",
  "messages": [
    {
      "role": "system",
      "content": "Du er en AI-assistent spesialisert i logisk tenkning og problemløsning. Brukeren vil delta i en flertrinns samtale med d[...]"
    },
    {
      "role": "user",
      "content": "I en liten by er det 1000 innbyggere. En ukjent andel av disse innbyggerne har en sjelden genetisk predisposisjon som gj[...]"
    },
    {
      "role": "assistant",
      "content": "Vi skal beregne den **Bayesianske faktoren (BF)** for nullhypotesen $ H_0: p = 0.1 $ mot alternativhypotesen $ H_1: p = [...]"
    }
  ],
  "source": "magpie-qwen3-bokmaal",
  "added": "2026-09-03",
  "created": "2025-05-01, 2025-05-31",
  "language": [
    "nob"
  ],
  "task": "reasoning",
  "token_count": 1362
}
```

### Data Fields

An entry in the dataset consists of the following fields:

- `id` (`str`): A unique identifier for each sample.
- `messages` (`list[dict]`): The conversation turns — see the [message format](../../README.md#data-fields) section in the main README.
- `source` (`str`): The source dataset (see [Source Data](#source-data)).
- `added` (`str`): The date when the sample was added to this collection.
- `created` (`str`): The approximate date range when the source data was created, as `"YYYY-MM-DD, YYYY-MM-DD"` (start, end). Year-level granularity is used when the exact date is unknown.
- `language` (`list[str]`): The language(s) of the sample as ISO 639-3 codes (e.g. `["nob"]`, `["nno"]`, `["nob", "eng"]`).
- `task` (`str`): The task type of the sample.
- `token_count` (`int`): The total number of tokens across all messages, computed using the Llama 3 tokenizer.
<!-- END-SAMPLE -->

### Dataset Statistics

<!-- START-DATASET PLOTS -->
<div style="display: flex; gap: 20px;">
<p align="center">
<img src="./images/dist_token_count.png" width="400" />
</p>
<p align="center">
<img src="./images/dist_turns.png" width="400" />
</p>
</div>
<!-- END-DATASET PLOTS -->

## Additional Information

### Source Data

The data comes from
[`versae/magpie-qwen3-235B-A22B-bokmaal`](https://huggingface.co/datasets/versae/magpie-qwen3-235B-A22B-bokmaal)
at revision `e47dc5c0f510425ec0ee064657467803e3477edf`. The source card says it was
generated in May 2025 with
[`Qwen3-235B-A22B`](https://huggingface.co/Qwen/Qwen3-235B-A22B), using the
[Magpie](https://arxiv.org/abs/2406.08464) method and no external seed data.

The `system`, `instruction`, and `response` fields become `system`, `user`, and
`assistant` messages. Empty system prompts are omitted, and the source category is used
as the task label.

### Quality Assurance

Both the instruction and response must be classified as Norwegian by `langid` 1.1.6.
This removes 177 of the 2,000 source rows. One duplicate conversation is also removed.

### Limitations

The source does not provide the exact generation code or inference settings. Because
the examples are synthetic, they may contain factual errors, unsafe material, or
incorrect task labels.

### License Information

The source dataset and Qwen3-235B-A22B are released under the
[Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0).