datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
slam-es-en-knowledge-tracing
SLAM Spanish-English Knowledge Tracing
This repository contains a Hugging Face-compatible conversion of the Spanish-English track from the 2018 Second Language Acquisition Modeling (SLAM) shared task.
Each row represents one learner exercise attempt. The original token-level mistake labels are retained in token_labels: 0 means OK and 1 means a mistake. The correct field is strict whole-attempt correctness: it is 1 only when every token label is 0.
This dataset is suitable for… See the full description on the dataset page: https://huggingface.co/datasets/bihungba1101/slam-es-en-knowledge-tracing.slam-en-es-knowledge-tracing
SLAM English-Spanish Knowledge Tracing
This repository contains a Hugging Face-compatible conversion of the English-Spanish track from the 2018 Second Language Acquisition Modeling (SLAM) shared task. Each row represents one learner exercise attempt. Source reference-token, linguistic, and official mistake-label annotations are preserved.
The dataset supports longitudinal knowledge tracing: the same anonymous learner can have many interactions over multiple relative days, with… See the full description on the dataset page: https://huggingface.co/datasets/bihungba1101/slam-en-es-knowledge-tracing.Knowledge-Tracing-QA-Chinese
🗂️ 数据集结构
数据集的每一行代表一条完整的专业问答/专家对话数据（标准 JSONL 格式），全量包含以下四个核心特征字段：
system (string): 动态系统提示词。明确锚定大模型在知识追踪、智能教育或数据挖掘领域的专家身份与回复规范。
instruction (string): 核心交互指令。包含知识追踪（KT）领域的模型评估方法、公式推导、算法对比等专业高难度问题。
input (string): 补充上下文或外部输入数据（如学生答题序列、习题交互矩阵）。若无可为空字符串。
output (string): 黄金标准回复。全量包含原生推理思维链（包裹在 <think>...</think> 标签内），随后输出极高质量的结构化学术级解答。
🧩 真实数据样例展现
以下为数据集中关于“因子模型遗忘能力评估”的真实单条数据切片（为保证阅读体验，部分极长正文已用 ... 省略）：
{
"system":… See the full description on the dataset page: https://huggingface.co/datasets/wuqi28/Knowledge-Tracing-QA-Chinese.