marin-community/open-thoughts-4-128-math-qwen3-30b-a3B-thinking-2507-annotated-32768-tokens

--- 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: 61807210 num_examples: 1024 download_size: 23156089 dataset_size: 61807210 configs: - config_name: default data_files: - split: train path: data/train-* --- # open-thoughts-4-128-math-qwen3-30b-a3B-thinking-2507-annotated-32768-tokens Math reasoning responses generated by **Qwen3-30B-A3B-Thinking-2507** ([Qwen/Qwen3-30B-A3B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-30B-A3B-Thinking-2507)). ## Overview - **Total rows:** 1,024 - **Unique prompts:** 128 (each with 8 response annotations) - **Source prompts:** [marin-community/open-thoughts-4-30k-math-qwen3-30b-a3B-thinking-2507-annotated-32768-tokens-n8-reformatted](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-30b-a3B-thinking-2507-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/Qwen3-30B-A3B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-30B-A3B-Thinking-2507) - **Max tokens:** 32,768 - **Temperature:** 0.8 - **Tokenizer used for stats:** Qwen/Qwen2.5-3B ## Statistics | Metric | Value | |--------|-------| | Avg tokens per response | 19,388 | | Median tokens per response | 17,645 | | Responses with `</think>` tag | 890/1024 (86.9%) | | Complete responses (has `</think>` + `\boxed{...}`) | 877/1024 (85.6%) | | Truncated responses | 147/1024 (14.4%) | | 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` | Qwen3-30B-A3B-Thinking-2507 generated response with a reasoning trace that typically ends with `</think>` | | `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 [model reasoning trace] </think> [final answer, typically containing \boxed{...}] ``` This model typically omits the opening `<think>` tag while still including the closing `</think>` tag. 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-qwen3-30b-a3B-thinking-2507-annotated-32768-tokens-n8-reformatted](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-30b-a3B-thinking-2507-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.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: 61807210 num_examples: 1024 download_size: 23156089 dataset_size: 61807210 configs: - config_name: default data_files: - split: train path: data/train-* # open-thoughts-4-128-math-qwen3-30b-a3B-thinking-2507-annotated-32768-tokens 由**Qwen3-30B-A3B-Thinking-2507**（[Qwen/Qwen3-30B-A3B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-30B-A3B-Thinking-2507)）生成的数学推理响应。 ## 概述 - **总样本数**：1,024 - **唯一提示数**：128（每个提示对应8条响应标注） - **源提示数据集**：[marin-community/open-thoughts-4-30k-math-qwen3-30b-a3B-thinking-2507-annotated-32768-tokens-n8-reformatted](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-30b-a3B-thinking-2507-annotated-32768-tokens-n8-reformatted) - **提示对齐**：与[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`完全匹配 - **生成模型**：Qwen3-30B-A3B-Thinking-2507（[Qwen/Qwen3-30B-A3B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-30B-A3B-Thinking-2507)） - **最大Token数**：32,768 - **温度系数**：0.8 - **用于统计的分词器**：Qwen/Qwen2.5-3B ## 统计指标 | 指标 | 数值 | |--------|-------| | 单条响应平均Token数 | 19,388 | | 单条响应Token数中位数 | 17,645 | | 包含`</think>`标签的响应 | 890/1024 (86.9%) | | 完整响应（同时包含`</think>`与`oxed{...}`） | 877/1024 (85.6%) | | 截断响应 | 147/1024 (14.4%) | | 空响应 | 0/1024 (0.0%) | ## 字段说明 | 字段 | 描述 | |--------|-------------| | `row_id` | 源自源数据集的行标识符 | | `instruction_seed` | 数学问题提示文本 | | `generated_text` | Qwen3-30B-A3B-Thinking-2507生成的响应，包含推理轨迹，通常以`</think>`结尾 | | `ms_id` | 数学种子ID，用于将同一提示的8条响应归为一组 | | `_source` | 源数据集标识符 | | `gpt41_mini_response` | GPT-4.1迷你版参考响应 | | `__original_row_idx` | 重格式化前的源数据流水线中的行索引 | | `length` | 源自源数据集的长度元数据 | | `final_answer` | 提取得到的最终答案（如存在） | | `complete_responses_count` | 该提示对应的源n=8组中完整响应的数量 | ## 响应格式 `generated_text`列中的每条响应通常遵循以下格式： text [模型推理轨迹] </think> [最终答案，通常包含oxed{...}] 该模型通常省略起始的`<think>`标签，但会包含闭合的`</think>`标签。 截断的响应可能缺少闭合的`</think>`标签和/或`oxed{...}`形式的答案。 ## 数据集构建 本数据集通过未经过打乱的方式，选取[marin-community/open-thoughts-4-30k-math-qwen3-30b-a3B-thinking-2507-annotated-32768-tokens-n8-reformatted](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-30b-a3B-thinking-2507-annotated-32768-tokens-n8-reformatted)的前1024行构建而成。 我们将`instruction_seed`序列与Kimi K2.5的128条提示参考数据集进行了比对，确认1024条样本的提示完全匹配。