marin-community/open-thoughts-4-128-math-qwen3-4b-annotated-32768-tokens
收藏Hugging Face2026-04-08 更新2026-04-12 收录
下载链接:
https://hf-mirror.com/datasets/marin-community/open-thoughts-4-128-math-qwen3-4b-annotated-32768-tokens
下载链接
链接失效反馈官方服务:
资源简介:
---
dataset_info:
features:
- name: row_id
dtype: int64
- name: instruction_seed
dtype: string
- name: _source
dtype: string
- name: gpt41_mini_response
dtype: string
- name: __original_row_idx
dtype: int64
- name: length
dtype: int64
- name: ms_id
dtype: int64
- name: generated_text
dtype: string
- name: final_answer
dtype: string
- name: complete_responses_count
dtype: int64
splits:
- name: train
num_bytes: 76226919
num_examples: 1024
download_size: 16120994
dataset_size: 76226919
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---
# open-thoughts-4-128-math-qwen3-4b-annotated-32768-tokens
Math reasoning responses generated by **Qwen3-4B** ([Qwen/Qwen3-4B](https://huggingface.co/Qwen/Qwen3-4B)).
## Overview
- **Total rows:** 1,024
- **Unique prompts:** 128 (each with 8 response annotations)
- **Source prompts:** [marin-community/open-thoughts-4-30k-math-qwen3-4b-annotated-32768-tokens-n8-reformatted](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-4b-annotated-32768-tokens-n8-reformatted)
- **Prompt alignment:** Exact `instruction_seed` match to [marin-community/open-thoughts-4-128-math-kimi-k2pt5-annotated-32768-tokens](https://huggingface.co/datasets/marin-community/open-thoughts-4-128-math-kimi-k2pt5-annotated-32768-tokens)
- **Generation model:** [Qwen/Qwen3-4B](https://huggingface.co/Qwen/Qwen3-4B)
- **Max tokens:** 32,768
- **Temperature:** 0.8
- **Tokenizer used for stats:** Qwen/Qwen2.5-3B
## Statistics
| Metric | Value |
|--------|-------|
| Avg tokens per response | 20,625 |
| Median tokens per response | 19,402 |
| Responses with `<think>` tag | 1024/1024 (100.0%) |
| Complete responses (has `</think>` + `\boxed{...}`) | 745/1024 (72.8%) |
| Truncated responses | 279/1024 (27.2%) |
| Empty responses | 0/1024 (0.0%) |
## Columns
| Column | Description |
|--------|-------------|
| `row_id` | Row identifier preserved from the source dataset |
| `instruction_seed` | The math problem prompt |
| `generated_text` | Qwen3-4B generated response with a `<think>...</think>` reasoning trace |
| `ms_id` | Math seed ID, groups all 8 responses for the same prompt |
| `_source` | Source dataset identifier |
| `gpt41_mini_response` | GPT-4.1 mini reference response |
| `__original_row_idx` | Row index from the pre-reformatted source pipeline |
| `length` | Length metadata carried over from the source dataset |
| `final_answer` | Extracted final answer when present |
| `complete_responses_count` | Number of complete responses in the source n=8 group for the prompt |
## Response Format
Each response in the `generated_text` column generally follows this format:
```text
<think>
[model reasoning trace]
</think>
[final answer, typically containing \boxed{...}]
```
This model emits an opening `<think>` tag for the reasoning trace.
Responses that are truncated may be missing the closing `</think>` tag and or the `\boxed{...}` answer.
## Construction
Created by taking the first 1,024 rows of [marin-community/open-thoughts-4-30k-math-qwen3-4b-annotated-32768-tokens-n8-reformatted](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-4b-annotated-32768-tokens-n8-reformatted) without shuffling.
The `instruction_seed` sequence was checked against the Kimi K2.5 128-prompt reference dataset and matched exactly across all 1,024 rows.
数据集信息:
特征:
- 字段名:row_id,数据类型:int64
- 字段名:instruction_seed,数据类型:字符串
- 字段名:_source,数据类型:字符串
- 字段名:gpt41_mini_response,数据类型:字符串
- 字段名:__original_row_idx,数据类型:int64
- 字段名:length,数据类型:int64
- 字段名:ms_id,数据类型:int64
- 字段名:generated_text,数据类型:字符串
- 字段名:final_answer,数据类型:字符串
- 字段名:complete_responses_count,数据类型:int64
数据划分:
- 划分名称:train,字节数:76226919,样本量:1024
下载大小:16120994
数据集总大小:76226919
配置项:
- 配置名称:default,数据文件:
- 划分:train,路径:data/train-*
# open-thoughts-4-128-math-qwen3-4b-annotated-32768-tokens
由通义千问3-4B(Qwen/Qwen3-4B)生成的数学推理回复,对应模型链接为[Qwen/Qwen3-4B](https://huggingface.co/Qwen/Qwen3-4B)。
## 数据集概览
- **总数据行数**:1024
- **唯一提示词数量**:128条(每条提示词对应8条回复标注)
- **源提示词数据集**:[marin-community/open-thoughts-4-30k-math-qwen3-4b-annotated-32768-tokens-n8-reformatted](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-4b-annotated-32768-tokens-n8-reformatted)
- **提示词对齐**:与数据集[marin-community/open-thoughts-4-128-math-kimi-k2pt5-annotated-32768-tokens](https://huggingface.co/datasets/marin-community/open-thoughts-4-128-math-kimi-k2pt5-annotated-32768-tokens)的`instruction_seed`字段完全匹配
- **生成模型**:通义千问3-4B(Qwen/Qwen3-4B),模型链接为[Qwen/Qwen3-4B](https://huggingface.co/Qwen/Qwen3-4B)
- **最大Token数**:32768
- **温度系数**:0.8
- **统计所用分词器**:Qwen/Qwen2.5-3B
## 统计指标
| 指标 | 数值 |
|------|------|
| 单回复平均Token数 | 20625 |
| 单回复Token数中位数 | 19402 |
| 包含`<think>`标签的回复 | 1024/1024(100.0%) |
| 完整回复(包含`</think>`与`oxed{...}`格式) | 745/1024(72.8%) |
| 截断回复 | 279/1024(27.2%) |
| 空回复 | 0/1024(0.0%) |
## 字段说明
| 字段名 | 说明 |
|-------|------|
| `row_id` | 源自源数据集的行标识符 |
| `instruction_seed` | 数学问题提示词 |
| `generated_text` | 通义千问3-4B生成的回复,包含`<think>...</think>`格式的推理过程 |
| `ms_id` | 数学提示词ID,用于将同一提示词对应的8条回复归为一组 |
| `_source` | 源数据集标识 |
| `gpt41_mini_response` | GPT-4.1 mini 参考回复 |
| `__original_row_idx` | 预格式化源数据流水线中的原始行索引 |
| `length` | 源自源数据集的长度元数据 |
| `final_answer` | 提取得到的最终答案(若存在) |
| `complete_responses_count` | 该提示词对应的源n=8分组中完整回复的数量 |
## 回复格式
`generated_text` 字段中的每条回复通常遵循以下格式:
text
<think>
[模型推理过程]
</think>
[最终答案,通常采用oxed{...}格式]
该模型会为推理过程添加起始`<think>`标签。截断的回复可能缺少闭合的`</think>`标签,或`oxed{...}`格式的最终答案。
## 数据集构建
本数据集通过截取数据集[marin-community/open-thoughts-4-30k-math-qwen3-4b-annotated-32768-tokens-n8-reformatted](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-4b-annotated-32768-tokens-n8-reformatted)的前1024行且未进行洗牌操作构建而成。
本数据集对`instruction_seed`字段与Kimi K2.5的128条提示词参考数据集进行了对齐校验,所有1024行的提示词均完全匹配。
提供机构:
marin-community


