{ "flow": { "nodes": [ { "id": "cmt-56258159441d", "type": "comment", "position": { "x": -800, "y": -200 }, "data": { "title": "Deep Reasoning: Logic Paradoxes & Probability", "text": "Six hard reasoning problems: Russell's paradox, Monty Hall, Blue Eyes, Unexpected Exam, Two-Envelope, Sleeping Beauty.\n\nThese require genuine deductive reasoning beyond pattern matching \u2014 they are unlikely to appear verbatim in training data.\n\nJudge scores: 5 = correct with full explanation, 1 = wrong or confabulated.", "color": "#fff5cc" } }, { "id": "tbl-02ee32294e1a", "type": "table", "position": { "x": -800, "y": 100 }, "data": { "title": "Logic & Probability Problems", "columns": [ { "header": "problem", "width": 160, "accessor": "c0" }, { "header": "question", "width": 380, "accessor": "c1" }, { "header": "expected", "width": 320, "accessor": "c2" } ], "rows": [ { "problem": "No barber shaves himself", "question": "A barber shaves all those, and only those, who do not shave themselves. Does the barber shave himself?", "expected": "paradox \u2014 no consistent answer exists (Russell's paradox)" }, { "problem": "Monty Hall", "question": "You pick door 1. Host opens door 3 (goat). Should you switch to door 2? Why?", "expected": "yes, switch \u2014 probability rises from 1/3 to 2/3" }, { "problem": "Blue eyes", "question": "100 logicians on an island all have blue eyes but don't know their own. A stranger says 'I see at least one person with blue eyes.' What happens and why?", "expected": "on day 100 all 100 leave \u2014 common knowledge trigger, inductive chain" }, { "problem": "Unexpected exam", "question": "A teacher says: 'There will be a surprise exam next week, you won't know the day until I announce it.' A student argues the exam is impossible. Is the student's logic correct?", "expected": "the student's backward induction is flawed \u2014 the paradox shows limits of purely deductive self-reference" }, { "problem": "Two-envelope", "question": "Two envelopes contain money; one has twice the other. You pick one and see \u00a3100. Should you switch? Formalize the paradox.", "expected": "the switch argument is flawed \u2014 expected value calculation assumes fixed reference point incorrectly; true expected value is equal" }, { "problem": "Sleeping Beauty", "question": "Sleeping Beauty is woken Monday always, and Monday+Tuesday if coin is tails. She has no memory. When woken, what probability should she assign to the coin being heads?", "expected": "halfer position: 1/2 (centered on coin flip); thirder position: 1/3 (centered on waking events) \u2014 both are defensible" } ] } }, { "id": "pmt-e8d25545fda2", "type": "prompt", "position": { "x": -150, "y": 100 }, "data": { "prompt": "Solve this problem carefully. Show your full reasoning.\n\n{question}", "llms": [ { "key": "hermes3-3a1df725f93e", "name": "hermes3", "emoji": "H", "model": "hermes3:latest", "base_model": "openai", "temp": 0.7, "settings": { "system_msg": "", "temperature": 0.7, "base_url": "http://localhost:11434/v1", "api_key": "ollama", "response_format": { "type": "text" }, "tools": [], "stop": [], "top_p": 1, "presence_penalty": 0, "frequency_penalty": 0 }, "formData": { "shortname": "hermes3", "model": "hermes3:latest", "system_msg": "", "temperature": 0.7, "base_url": "http://localhost:11434/v1", "api_key": "ollama", "response_format": "text", "tools": "", "stop": "", "top_p": 1, "presence_penalty": 0, "frequency_penalty": 0 } }, { "key": "qwen3-46d891b4466e", "name": "qwen3", "emoji": "Q", "model": "qwen3:14b", "base_model": "openai", "temp": 0.7, "settings": { "system_msg": "", "temperature": 0.7, "base_url": "http://localhost:11434/v1", "api_key": "ollama", "response_format": { "type": "text" }, "tools": [], "stop": [], "top_p": 1, "presence_penalty": 0, "frequency_penalty": 0 }, "formData": { "shortname": "qwen3", "model": "qwen3:14b", "system_msg": "", "temperature": 0.7, "base_url": "http://localhost:11434/v1", "api_key": "ollama", "response_format": "text", "tools": "", "stop": "", "top_p": 1, "presence_penalty": 0, "frequency_penalty": 0 } }, { "key": "phi4-0da188d36729", "name": "phi4", "emoji": "P", "model": "phi4:latest", "base_model": "openai", "temp": 0.7, "settings": { "system_msg": "", "temperature": 0.7, "base_url": "http://localhost:11434/v1", "api_key": "ollama", "response_format": { "type": "text" }, "tools": [], "stop": [], "top_p": 1, "presence_penalty": 0, "frequency_penalty": 0 }, "formData": { "shortname": "phi4", "model": "phi4:latest", "system_msg": "", "temperature": 0.7, "base_url": "http://localhost:11434/v1", "api_key": "ollama", "response_format": "text", "tools": "", "stop": "", "top_p": 1, "presence_penalty": 0, "frequency_penalty": 0 } }, { "key": "llama3.1-46b65e85729a", "name": "llama3.1", "emoji": "L", "model": "llama3.1:8b", "base_model": "openai", "temp": 0.7, "settings": { "system_msg": "", "temperature": 0.7, "base_url": "http://localhost:11434/v1", "api_key": "ollama", "response_format": { "type": "text" }, "tools": [], "stop": [], "top_p": 1, "presence_penalty": 0, "frequency_penalty": 0 }, "formData": { "shortname": "llama3.1", "model": "llama3.1:8b", "system_msg": "", "temperature": 0.7, "base_url": "http://localhost:11434/v1", "api_key": "ollama", "response_format": "text", "tools": "", "stop": "", "top_p": 1, "presence_penalty": 0, "frequency_penalty": 0 } } ], "n": 1 } }, { "id": "eval-98926ecbd228", "type": "llmeval", "position": { "x": 420, "y": 100 }, "data": { "title": "Reasoning Judge (1-5)", "grader": { "key": "judge-b8244427cc66", "name": "llama3.1 (judge)", "emoji": "J", "model": "llama3.1:8b", "base_model": "openai", "temp": 0.1, "settings": { "system_msg": "", "temperature": 0.1, "base_url": "http://localhost:11434/v1", "api_key": "ollama", "response_format": { "type": "text" }, "tools": [], "stop": [] }, "formData": { "shortname": "llama3.1", "model": "llama3.1:8b", "system_msg": "", "temperature": 0.1, "base_url": "http://localhost:11434/v1", "api_key": "ollama" } }, "prompt": "Problem: {problem}\nExpected answer: {expected}\n\nModel response:\n{response}\n\nRate reasoning quality 1-5:\n5 = correct answer with sound logical justification\n4 = correct answer with minor gaps in reasoning\n3 = partially correct, some reasoning shown\n2 = wrong answer but shows some relevant reasoning\n1 = wrong answer with no valid reasoning or confabulated\n\nReply with only a single integer 1-5." } }, { "id": "vis-65091f31949b", "type": "vis", "position": { "x": 900, "y": 100 }, "data": {} } ], "edges": [ { "id": "e-5c43033b1b96-question", "source": "tbl-02ee32294e1a", "sourceHandle": "question", "target": "pmt-e8d25545fda2", "targetHandle": "question", "markerEnd": { "type": "arrowclosed", "width": 22, "height": 22 } }, { "id": "e-d318f2b2a95c", "source": "pmt-e8d25545fda2", "sourceHandle": "prompt", "target": "eval-98926ecbd228", "targetHandle": "responseBatch", "markerEnd": { "type": "arrowclosed", "width": 22, "height": 22 } }, { "id": "e-1a14d4667f42", "source": "eval-98926ecbd228", "sourceHandle": "output", "target": "vis-65091f31949b", "targetHandle": "input", "markerEnd": { "type": "arrowclosed", "width": 22, "height": 22 } } ] } }