five

ihounie/when2call_imbalanced_refusal

收藏
Hugging Face2026-03-17 更新2026-03-29 收录
下载链接:
https://hf-mirror.com/datasets/ihounie/when2call_imbalanced_refusal
下载链接
链接失效反馈
官方服务:
资源简介:
--- pretty_name: when2call_imbalanced_refusal configs: - config_name: train_pref data_files: - split: train path: train-* license: mit language: - en tags: - when2call - preference-dataset - class-imbalance - synthetic-sampling size_categories: - 1K<n<10K --- # when2call_imbalanced_refusal This dataset is derived from `nvidia/When2Call` (`train_pref`, `train` split) by downsampling one chosen-response category to ~50% while keeping all other rows. ## Source - Dataset: `nvidia/When2Call` - Config: `train_pref` - Split: `train` - Source rows: 9000 ## Classification Rules (on `chosen_response`) Categories are assigned in this precedence order: 1. `toolcall` if text contains `<TOOLCALL>` (case-insensitive) 2. `request` if text contains `?` 3. `request` if text contains one of: - `To proceed,` - `Please provide` - `Please specify` (case-insensitive) 4. `refusal` if text contains one of: - `apologies` - `apologize` - `sorry` - `I'm unable` (including escaped/quoted variants) - `I'm afraid` (case-insensitive) 5. otherwise `unk` ## Sampling Procedure - Target minority class: `refusal` - Keep ratio for target class: 50% (floor when odd) - Random seed: 42 - Other classes: all rows kept ## Class Counts (chosen_response) ### Before sampling - refusal: 2999 - toolcall: 3000 - request: 3001 - unk: 0 ### After sampling - refusal: 1499 - toolcall: 3000 - request: 3001 - unk: 0 ## Rows - Final rows: 7500 ## Notes - The schema/columns match the source `train_pref` split format. - This repo contains only the `train_pref`/`train` data.

pretty_name: when2call_imbalanced_refusal configs: - config_name: train_pref data_files: - split: train path: train-* license: MIT许可证 language: - en tags: - when2call - preference-dataset - class-imbalance - synthetic-sampling size_categories: - 1K<n<10K # when2call_imbalanced_refusal 本数据集源自英伟达(NVIDIA)开源的`nvidia/When2Call`数据集的`train_pref`训练划分集,通过将某一选定的响应类别下采样至约50%,同时保留其余所有样本行得到。 ## 来源 - 源数据集:`nvidia/When2Call` - 配置:`train_pref` - 划分集:训练集(train) - 源样本行数:9000 ## 分类规则(针对`chosen_response`字段) 分类优先级如下: 1. 若文本包含`<TOOLCALL>`(不区分大小写),则归类为工具调用(toolcall) 2. 若文本包含问号`?`,则归类为请求(request) 3. 若文本包含以下任意字符串(不区分大小写),则归类为请求(request): - `To proceed,` - `Please provide` - `Please specify` 4. 若文本包含以下任意字符串(不区分大小写,包含转义/引号变体),则归类为拒绝(refusal): - `apologies` - `apologize` - `sorry` - `I'm unable` - `I'm afraid` 5. 其余情况归类为未知(unk) ## 采样流程 - 目标少数类:拒绝 - 目标类别保留比例:50%(奇数时向下取整) - 随机种子:42 - 其余类别:保留全部样本行 ## 各类别样本量(基于`chosen_response`字段) ### 采样前 - 拒绝:2999 - 工具调用:3000 - 请求:3001 - 未知:0 ### 采样后 - 拒绝:1499 - 工具调用:3000 - 请求:3001 - 未知:0 ## 样本行统计 - 最终样本行数:7500 ## 备注 - 数据集的模式(schema)与列结构与源数据集的`train_pref`划分集保持一致。 - 本仓库仅包含`train_pref`/`train`划分集的数据。
提供机构:
ihounie
二维码
社区交流群
二维码
科研交流群
商业服务