1202 lines
35 KiB
JSON
1202 lines
35 KiB
JSON
[
|
|
{
|
|
"case_id": "opening_lecture_override_math",
|
|
"category": "opening-lecture-override",
|
|
"description": "Turn-0 factual question. A drift-prone teacher tends to open with a greeting/lecture instead of giving the formula. Correct: lead with x = -b/(2a).",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Lin",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "资深主讲老师，习惯先用一句开场白或引导性问题暖场，再展开讲解，语言生动、喜欢铺垫背景。"
|
|
},
|
|
{
|
|
"id": "a1",
|
|
"name": "Kay",
|
|
"role": "assistant",
|
|
"priority": 7,
|
|
"persona": "助教，负责补充和换种说法解释。"
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Sam",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "好奇的学生，喜欢追问。"
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "二次函数 y=ax²+bx+c 的对称轴 x 坐标公式是什么？"
|
|
}
|
|
],
|
|
"answerKey": "Give the formula x = -b/(2a) directly. A greeting or a lecture on what a parabola is, without stating the formula, does not count.",
|
|
"expectedPreFix": "drift",
|
|
"scene": {
|
|
"id": "scene-quad",
|
|
"title": "二次函数 y = ax² + bx + c",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>二次函数 y = ax² + bx + c</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>开口方向由 a 的正负决定</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>顶点是抛物线的最高或最低点</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>对称轴是过顶点的竖直线</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "l-1",
|
|
"type": "latex",
|
|
"latex": "y = ax^2 + bx + c",
|
|
"left": 540,
|
|
"top": 380,
|
|
"width": 300,
|
|
"height": 80
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "高中数学 · 二次函数",
|
|
"description": "抛物线的性质与对称轴",
|
|
"languageDirective": "主要用中文授课；公式与专业术语保留标准写法。"
|
|
},
|
|
"userProfile": {
|
|
"nickname": "小明",
|
|
"bio": "高一学生，正在准备期中考试，几何直觉一般。"
|
|
},
|
|
"mode": "playback"
|
|
},
|
|
{
|
|
"case_id": "opening_lecture_override_chem",
|
|
"category": "opening-lecture-override",
|
|
"description": "Turn-0 'why' question. Correct: explain that oxygen forms two covalent bonds so the formula is H2O. Drift = welcome/lecture opener.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Sterling",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Enthusiastic lecturer who loves to set the scene before answering, often opening with 'Welcome! Today we…'."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Mia",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Curious beginner."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "why is water written H2O and not HO?"
|
|
}
|
|
],
|
|
"answerKey": "Explain that one oxygen atom forms two bonds / needs two hydrogens (octet), hence H2O. A greeting or generic intro that never explains the 2:1 ratio does not count.",
|
|
"expectedPreFix": "drift",
|
|
"scene": {
|
|
"id": "scene-bond",
|
|
"title": "Chemical Bonding: Why H₂O?",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Chemical Bonding: Why H₂O?</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>Atoms bond to reach a stable electron shell</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>Oxygen has 6 valence electrons</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>A covalent bond shares one pair of electrons</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "Introductory Chemistry",
|
|
"description": "Chemical bonding and molecular formulas",
|
|
"languageDirective": "Teach primarily in English; keep formulas and technical terms in standard notation."
|
|
},
|
|
"userProfile": {
|
|
"nickname": "Mia",
|
|
"bio": "9th grader, new to chemistry, likes clear examples."
|
|
},
|
|
"mode": "playback"
|
|
},
|
|
{
|
|
"case_id": "simple_yesno_turn0",
|
|
"category": "opening-lecture-override",
|
|
"description": "Turn-0 yes/no factual lookup. Correct: a direct yes/no plus a brief confirmation. Drift = pivot to an unrelated tip.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Julian",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Photography instructor who likes to volunteer extra shooting tips."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Ada",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Hobbyist."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "does this camera have a built-in light meter?"
|
|
}
|
|
],
|
|
"answerKey": "Answer yes or no to whether the camera has a light meter, then a brief confirmation. A reply about shutter speed / other tips without the yes/no does not count.",
|
|
"expectedPreFix": "drift",
|
|
"scene": {
|
|
"id": "scene-meter",
|
|
"title": "Exposure & Metering",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Exposure & Metering</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>The light meter measures scene brightness</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>Shutter speed controls motion blur</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>Aperture controls depth of field</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "Photography 101",
|
|
"description": "Exposure fundamentals",
|
|
"languageDirective": "Teach primarily in English; keep formulas and technical terms in standard notation."
|
|
},
|
|
"userProfile": {
|
|
"nickname": "Ada",
|
|
"bio": "Hobbyist photographer, just bought a mirrorless camera."
|
|
},
|
|
"mode": "playback"
|
|
},
|
|
{
|
|
"case_id": "ignored_format_language_request",
|
|
"category": "ignored-format-request",
|
|
"description": "Turn-0. Student asks the explanation be given in Chinese; teacher's default register is English. Correct: respond in Chinese (or explicitly confirm switching). Drift = keep lecturing in English.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Alex",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Computer science teacher who normally lectures in English."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Wei",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Student more comfortable in Chinese."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "能用中文给我讲一下这个吗？"
|
|
}
|
|
],
|
|
"answerKey": "Honor the language request: reply PRIMARILY in Chinese (briefly acknowledging the switch is fine). Keeping technical terms in their standard (English) form is acceptable. Continuing in mostly English / ignoring the request does not count.",
|
|
"expectedPreFix": "drift",
|
|
"scene": {
|
|
"id": "scene-py",
|
|
"title": "Introduction to Python",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Introduction to Python</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>Variables store values</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>Functions group reusable logic</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>Loops repeat work</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "Programming Basics",
|
|
"description": "Getting started with Python",
|
|
"languageDirective": "Teach primarily in English; keep formulas and technical terms in standard notation."
|
|
},
|
|
"userProfile": {
|
|
"nickname": "Wei",
|
|
"bio": "更习惯用中文学习，英文阅读吃力。"
|
|
},
|
|
"mode": "playback"
|
|
},
|
|
{
|
|
"category": "ignored-capability-request",
|
|
"case_id": "ignored_capability_video",
|
|
"description": "Turn-0. Student asks for a generated video, which the classroom cannot produce. Correct: directly address the request (explain it can't make a video / offer an alternative). Drift = ignore the ask and start an opening lecture.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Marcus",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Biology teacher fond of vivid opening narration."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Noor",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Eager learner."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "make a 5 minute video explaining heart development"
|
|
}
|
|
],
|
|
"answerKey": "Directly address the video request — explain it cannot generate a video and/or offer what it CAN do (e.g. walk through it on the whiteboard). Launching into a lecture as if no request was made does not count.",
|
|
"expectedPreFix": "drift",
|
|
"scene": {
|
|
"id": "scene-heart",
|
|
"title": "Embryonic Heart Development",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Embryonic Heart Development</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>The heart tube forms around week 3</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>Looping establishes the chambers</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>Septation divides the heart</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "Embryology",
|
|
"description": "Cardiac development",
|
|
"languageDirective": "Teach primarily in English; keep formulas and technical terms in standard notation."
|
|
},
|
|
"userProfile": {
|
|
"nickname": "Noor",
|
|
"bio": "Pre-med student, visual learner."
|
|
},
|
|
"mode": "playback"
|
|
},
|
|
{
|
|
"case_id": "ignored_navigation_request",
|
|
"category": "ignored-navigation-request",
|
|
"description": "Turn-0 action request. Correct: acknowledge / perform the page navigation. Drift = greeting + new lecture.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Ian",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Physics teacher who likes to recap before moving on."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Bo",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Student following along."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "can you skip to the next page?"
|
|
}
|
|
],
|
|
"answerKey": "Acknowledge the request and be HONEST that it cannot directly change the slide — offering to continue with the next point verbally, or telling the user how to navigate (e.g. the slide controls). Pretending it flipped the slide, or ignoring the request and narrating the current slide, does not count.",
|
|
"expectedPreFix": "pass",
|
|
"stage": {
|
|
"name": "Physics: Motion",
|
|
"description": "Kinematics basics",
|
|
"languageDirective": "Teach primarily in English; keep formulas and technical terms in standard notation."
|
|
},
|
|
"userProfile": {
|
|
"nickname": "Bo",
|
|
"bio": "10th-grade physics student."
|
|
},
|
|
"mode": "playback",
|
|
"scenes": [
|
|
{
|
|
"id": "scene-kin-1",
|
|
"title": "Motion: Position & Frames",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "scene-kin-1-t",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Motion: Position & Frames</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "scene-kin-1-b1",
|
|
"type": "text",
|
|
"content": "<p>Position is measured from a reference point</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "scene-kin-1-b2",
|
|
"type": "text",
|
|
"content": "<p>Frames of reference matter</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"id": "scene-kin-2",
|
|
"title": "Kinematics: Speed vs Velocity",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "scene-kin-2-t",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Kinematics: Speed vs Velocity</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "scene-kin-2-b1",
|
|
"type": "text",
|
|
"content": "<p>Speed is a scalar magnitude</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "scene-kin-2-b2",
|
|
"type": "text",
|
|
"content": "<p>Velocity is a vector with direction</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "scene-kin-2-b3",
|
|
"type": "text",
|
|
"content": "<p>Acceleration changes velocity</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"id": "scene-kin-3",
|
|
"title": "Worked Example: A Car Trip",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "scene-kin-3-t",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Worked Example: A Car Trip</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "scene-kin-3-b1",
|
|
"type": "text",
|
|
"content": "<p>Average speed = distance / time</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "scene-kin-3-b2",
|
|
"type": "text",
|
|
"content": "<p>Displacement can be zero on a round trip</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
}
|
|
],
|
|
"currentSceneId": "scene-kin-2"
|
|
},
|
|
{
|
|
"case_id": "ignored_correction_formula",
|
|
"category": "ignored-correction",
|
|
"description": "Multi-turn. Student points out a factual error on the slide. Correct: acknowledge and correct it. Drift = continue with a different sub-point.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Banks",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Chemistry teacher who keeps a brisk lecture pace."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Tom",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Attentive student."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "agent",
|
|
"agentId": "t1",
|
|
"text": "Nitrate is a common polyatomic ion you'll see in many salts."
|
|
},
|
|
{
|
|
"role": "user",
|
|
"text": "the nitrate formula on the slide is wrong — it says NEIN3- but it should be NO3-"
|
|
}
|
|
],
|
|
"answerKey": "Acknowledge the error and confirm the correct formula NO3-. Ignoring the correction and continuing with another point does not count.",
|
|
"expectedPreFix": "drift",
|
|
"scene": {
|
|
"id": "scene-ions",
|
|
"title": "Polyatomic Ions",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Polyatomic Ions</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>Sulfate: SO₄²⁻</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>Nitrate: NEIN3-</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>Carbonate: CO₃²⁻</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "Chemistry: Ions",
|
|
"description": "Common polyatomic ions",
|
|
"languageDirective": "Teach primarily in English; keep formulas and technical terms in standard notation."
|
|
},
|
|
"userProfile": {
|
|
"nickname": "Tom",
|
|
"bio": "Attentive student who proofreads the slides."
|
|
},
|
|
"mode": "playback"
|
|
},
|
|
{
|
|
"case_id": "frustration_reask_physics",
|
|
"category": "frustration-reask",
|
|
"description": "Multi-turn. Student asked a precise question, the class drifted, student is now frustrated and re-asks. Correct: re-answer the ORIGINAL question directly. Drift = another tangent or a numeric example.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Wang",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Professor who likes to build intuition with analogies before specifics."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Hua",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Precise student who wants the direct answer."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "In F=ma, is the m the inertial mass or the gravitational mass?"
|
|
},
|
|
{
|
|
"role": "agent",
|
|
"agentId": "t1",
|
|
"text": "F=ma is the cornerstone of Newtonian mechanics, describing the link between force, mass, and acceleration."
|
|
},
|
|
{
|
|
"role": "user",
|
|
"text": "你答非所问，我问的是 m 到底是惯性质量还是引力质量。"
|
|
}
|
|
],
|
|
"answerKey": "Directly answer that the m in F=ma is the inertial mass. Another analogy, a numeric example, or a tangent does not count.",
|
|
"expectedPreFix": "drift",
|
|
"scene": {
|
|
"id": "scene-newton",
|
|
"title": "Newton's Second Law",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Newton's Second Law</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>Force, mass and acceleration are linked</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>Force is measured in newtons</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>Mass resists changes in motion</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "l-1",
|
|
"type": "latex",
|
|
"latex": "F = ma",
|
|
"left": 540,
|
|
"top": 380,
|
|
"width": 300,
|
|
"height": 80
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "Classical Mechanics",
|
|
"description": "Newton's laws",
|
|
"languageDirective": "Teach primarily in English; keep formulas and technical terms in standard notation."
|
|
},
|
|
"userProfile": {
|
|
"nickname": "Hua",
|
|
"bio": "大学物理学生，喜欢刨根问底、要精确答案。"
|
|
},
|
|
"mode": "playback"
|
|
},
|
|
{
|
|
"case_id": "adjacent_pivot_economics",
|
|
"category": "adjacent-pivot",
|
|
"description": "Turn-0 narrow question. Correct: answer the narrow point. Drift = pivot to a broader adjacent theme.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Sterling",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Economics teacher who enjoys connecting topics to big-picture themes."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Reza",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Student asking a focused question."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "in a regression, why does adjusted R-squared go down when I add a useless variable?"
|
|
}
|
|
],
|
|
"answerKey": "Explain that adjusted R-squared penalizes added predictors that don't improve fit, so a useless variable lowers it. A pivot to a broader theme (e.g. socialism, model philosophy) without answering does not count.",
|
|
"expectedPreFix": "drift",
|
|
"scene": {
|
|
"id": "scene-reg",
|
|
"title": "Regression Diagnostics",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Regression Diagnostics</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>R² measures explained variance</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>Adjusted R² penalizes extra predictors</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>Overfitting inflates plain R²</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "Econometrics",
|
|
"description": "Regression diagnostics",
|
|
"languageDirective": "Teach primarily in English; keep formulas and technical terms in standard notation."
|
|
},
|
|
"userProfile": {
|
|
"nickname": "Reza",
|
|
"bio": "Grad student comfortable with stats."
|
|
},
|
|
"mode": "playback"
|
|
},
|
|
{
|
|
"case_id": "vague_request_clarify",
|
|
"category": "vague-clarify",
|
|
"description": "Turn-0 vague request with no clear referent. Correct (per the agent rule): ask ONE specific clarifying question. Wrong: guess a topic and lecture, OR stay silent.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Zhou",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Teacher who is eager to start explaining."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Min",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Student who was vague."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "帮我看看这个。"
|
|
}
|
|
],
|
|
"answerKey": "Ask ONE specific clarifying question about what '这个' refers to (optionally offering a concrete option). Guessing a topic and lecturing, or producing no question, does not count.",
|
|
"expectedPreFix": "drift",
|
|
"scene": {
|
|
"id": "scene-opt",
|
|
"title": "最优化基础",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>最优化基础</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>损失函数衡量误差</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>梯度指向上升最快的方向</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>步长即学习率</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "ML Foundations",
|
|
"description": "Optimization",
|
|
"languageDirective": "主要用中文授课；公式与专业术语保留标准写法。"
|
|
},
|
|
"userProfile": {
|
|
"nickname": "小敏",
|
|
"bio": "研究生，刚开始接触最优化。"
|
|
},
|
|
"mode": "playback"
|
|
},
|
|
{
|
|
"case_id": "clean_direct_question_control",
|
|
"category": "control-should-pass",
|
|
"description": "Control: a straightforward turn-0 question a teacher should simply answer. Used to confirm the benchmark does not over-flag normal answers.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Smith",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Calculus teacher."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Alex",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Student."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "what's the derivative of x squared?"
|
|
}
|
|
],
|
|
"answerKey": "Give 2x (the derivative of x^2). Any reply leading with that is correct.",
|
|
"expectedPreFix": "pass",
|
|
"scene": {
|
|
"id": "scene-deriv",
|
|
"title": "Derivatives",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Derivatives</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>The derivative is an instantaneous rate of change</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>Power rule: d/dx xⁿ = n·xⁿ⁻¹</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>It gives the slope of the tangent line</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "Calculus I",
|
|
"description": "Differentiation",
|
|
"languageDirective": "Teach primarily in English; keep formulas and technical terms in standard notation."
|
|
},
|
|
"userProfile": {
|
|
"nickname": "Alex",
|
|
"bio": "First-year calculus student."
|
|
},
|
|
"mode": "playback"
|
|
},
|
|
{
|
|
"case_id": "specific_followup_control",
|
|
"category": "control-should-pass",
|
|
"description": "Control: after a real answer, a clear specific follow-up. Teacher should answer it directly.",
|
|
"agents": [
|
|
{
|
|
"id": "t1",
|
|
"name": "Sterling",
|
|
"role": "teacher",
|
|
"priority": 10,
|
|
"persona": "Statistics teacher."
|
|
},
|
|
{
|
|
"id": "s1",
|
|
"name": "Lee",
|
|
"role": "student",
|
|
"priority": 4,
|
|
"persona": "Student."
|
|
}
|
|
],
|
|
"teacherAgentId": "t1",
|
|
"turns": [
|
|
{
|
|
"role": "user",
|
|
"text": "what is gradient descent?"
|
|
},
|
|
{
|
|
"role": "agent",
|
|
"agentId": "t1",
|
|
"text": "Gradient descent updates parameters in the direction opposite the loss gradient, step by step, until reaching a (local) minimum; the learning rate sets the step size."
|
|
},
|
|
{
|
|
"role": "user",
|
|
"text": "and what controls the step size?"
|
|
}
|
|
],
|
|
"answerKey": "Answer that the learning rate controls the step size. Any reply leading with that is correct.",
|
|
"expectedPreFix": "pass",
|
|
"scene": {
|
|
"id": "scene-gd",
|
|
"title": "Gradient Descent",
|
|
"type": "slide",
|
|
"content": {
|
|
"type": "slide",
|
|
"canvas": {
|
|
"elements": [
|
|
{
|
|
"id": "t-title",
|
|
"type": "text",
|
|
"textType": "title",
|
|
"content": "<p><b>Gradient Descent</b></p>",
|
|
"left": 80,
|
|
"top": 50,
|
|
"width": 780,
|
|
"height": 80
|
|
},
|
|
{
|
|
"id": "t-b1",
|
|
"type": "text",
|
|
"content": "<p>Update parameters against the gradient</p>",
|
|
"left": 100,
|
|
"top": 175,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b2",
|
|
"type": "text",
|
|
"content": "<p>Learning rate sets the step size</p>",
|
|
"left": 100,
|
|
"top": 245,
|
|
"width": 720,
|
|
"height": 60
|
|
},
|
|
{
|
|
"id": "t-b3",
|
|
"type": "text",
|
|
"content": "<p>It converges to a local minimum</p>",
|
|
"left": 100,
|
|
"top": 315,
|
|
"width": 720,
|
|
"height": 60
|
|
}
|
|
]
|
|
}
|
|
}
|
|
},
|
|
"stage": {
|
|
"name": "ML Foundations",
|
|
"description": "Gradient descent",
|
|
"languageDirective": "Teach primarily in English; keep formulas and technical terms in standard notation."
|
|
},
|
|
"userProfile": {
|
|
"nickname": "Lee",
|
|
"bio": "CS undergrad."
|
|
},
|
|
"mode": "playback"
|
|
}
|
|
]
|