conversations listlengths 4 4 |
|---|
[
{
"from": "human",
"value": "what models are you?"
},
{
"from": "gpt",
"value": "<think>\nwhat models are you?\n\nWe must follow policies. The user asks what models are you? We can respond about identity. We should not hallucinate. We are AI assistant called AI assistant. Probably we can say we ... |
[
{
"from": "human",
"value": "Rarest mineral"
},
{
"from": "gpt",
"value": "<think>\nRarest mineral\n\nWe need to respond according to policy: never fabricate real-world facts... No access to search engines. Can't verify rarest mineral. Must explain can't verify. Could give general landscape.\n\n... |
[
{
"from": "human",
"value": "Test"
},
{
"from": "gpt",
"value": "<think>\nTest\n\nWe need to respond to test. Probably just acknowledge. No policy violation. Could say hello.\n\nFollow policy. Simple.\n\nProbably respond with greeting.\n\nWe must maintain honesty policy. No issues.\n\nUser says ... |
[
{
"from": "human",
"value": "Site: \"m.egwwritings.org/en\" über die flache Erde"
},
{
"from": "gpt",
"value": "<think>\nSite: \"m.egwwritings.org/en\" über die flache Erde\n\nSite: \"m.egwwritings.org/en\" about flat earth\n\nWe need to respond in German presumably. User asks about site m.egwwr... |
[
{
"from": "human",
"value": "Ellen White über die Flache Erde"
},
{
"from": "gpt",
"value": "<think>\nEllen White über die Flache Erde\n\nWe need respond in German likely. User asks about Ellen White about flat earth. Must follow policies. We cannot hallucinate. Ellen White was Seventh-day Adven... |
[
{
"from": "human",
"value": "تجم تفسرلي علاش فما صبع زايد في مرض blackfond diamond ?"
},
{
"from": "gpt",
"value": "<think>\nتجم تفسرلي علاش فما صبع زايد في مرض blackfond diamond ?\n\nInterpretation: Tunisian dialect? \"Can you explain to me why there is an extra finger in black fond diamond dis... |
[
{
"from": "human",
"value": "find me where and when is trump doing speech today after talk with putin"
},
{
"from": "gpt",
"value": "<think>\nfind me where and when is trump doing speech today after talk with putin\n\nWe need to follow policy: never fabricate real-world facts, statistics, dates,... |
[
{
"from": "human",
"value": "What's the weather in Sf today"
},
{
"from": "gpt",
"value": "<think>\nWhat's the weather in Sf today\n\nWe need to follow policy: question about current events, today's news, live prices, sports results, who currently holds some position, or anything time-sensitive ... |
[
{
"from": "human",
"value": "日本はどこですか"
},
{
"from": "gpt",
"value": "<think>\n日本はどこですか\n\nJapanese: Where is Japan?\n\nWe need respond in Japanese presumably. The policy says never fabricate real-world facts... Where is Japan is verifiable general knowledge. That's fine. It's geographic. No need... |
[
{
"from": "human",
"value": "Are there two characters in the IDW comics that have been confirmed to be LGBTQIA+?"
},
{
"from": "gpt",
"value": "<think>\nAre there two characters in the IDW comics that have been confirmed to be LGBTQIA+?\n\nWe need to follow policy. Question about current events?... |
[
{
"from": "human",
"value": "Are Tangle and Whisper from Sonic gay"
},
{
"from": "gpt",
"value": "<think>\nAre Tangle and Whisper from Sonic gay\n\nWe need to respond per policy: never fabricate real-world facts... We have no search. Tangle and Whisper from Sonic gay? This is about shipping? We ... |
[
{
"from": "human",
"value": "كيف قرأ قالون كلمت سيئت"
},
{
"from": "gpt",
"value": "<think>\nكيف قرأ قالون كلمت سيئت\n\nHow did Qalun read the word \"سيئت\"? Need answer about Quranic reading.\n\nUser asks in Arabic: How did Qalan read the word si't? سيئت\n\nLikely refers to Quranic verse? سيئت ... |
[
{
"from": "human",
"value": "Can you tell me the latest updates of a cat named \"Tomay\" via Instagram?"
},
{
"from": "gpt",
"value": "<think>\nCan you tell me the latest updates of a cat named \"Tomay\" via Instagram?\n\nWe need to follow policies: question about current events, today's news, l... |
[
{
"from": "human",
"value": "Hi"
},
{
"from": "gpt",
"value": "<think>\nThe user has provided a simple greeting (\"Hi\"). This is a standard opening to a conversation. My goal is to acknowledge the greeting politely and offer assistance. Since there is no specific question or task yet, I should ... |
[
{
"from": "human",
"value": "what day does long bridge have meatball subs"
},
{
"from": "gpt",
"value": "<think>\nwhat day does long bridge have meatball subs\n\nWe need to follow policy. This is a real-world fact about a restaurant menu. Possibly Long Bridge? Long Bridge Pizza? Long Bridge... C... |
[
{
"from": "human",
"value": "how did Hajduk play its last match"
},
{
"from": "gpt",
"value": "<think>\nhow did Hajduk play its last match\n\nWe need to answer about Hajduk Split last match. Need current? The user asks how did Hajduk play its last match. That's real-world fact that could have ch... |
[
{
"from": "human",
"value": "Could you research on how to evaluate Agentic AI applications ?"
},
{
"from": "gpt",
"value": "<think>\nCould you research on how to evaluate Agentic AI applications ?\n\nWe need to follow policies: never fabricate real-world facts, statistics, dates, prices, names, ... |
[
{
"from": "human",
"value": "When does train 405 caltrain reach san mateo"
},
{
"from": "gpt",
"value": "<think>\nWhen does train 405 caltrain reach san mateo\n\nWe need to follow policies. This is time-sensitive, current schedule. We cannot reliably answer. No live search. Must say we cannot re... |
[
{
"from": "human",
"value": "Are there two characters in the IDW Sonic comics that have been confirmed to be LGBTQIA+?"
},
{
"from": "gpt",
"value": "<think>\nAre there two characters in the IDW Sonic comics that have been confirmed to be LGBTQIA+?\n\nWe need to answer about IDW Sonic comics LGB... |
[
{
"from": "human",
"value": "what time does caltrain 405 reach san mateo"
},
{
"from": "gpt",
"value": "<think>\nwhat time does caltrain 405 reach san mateo\n\nWe need to follow policy: Do not hallucinate real-world facts. If answer depends on something that could have changed... must not confid... |
[
{
"from": "human",
"value": "كيف قرأ حمزة كلمة الأرض"
},
{
"from": "gpt",
"value": "<think>\nكيف قرأ حمزة كلمة الأرض\n\nHow did Hamza read the word الأرض?\n\nWe need answer in Arabic. How did Hamza read word الأرض.\n\nIn Quranic reading, Hamza's recitation? Hamza al-Zayyat? The word الأرض.\n\nTh... |
[
{
"from": "human",
"value": "Tell me all about NVDA GTC"
},
{
"from": "gpt",
"value": "<think>\nTell me all about NVDA GTC\n\nWe need to respond according to policy. User asks about NVDA GTC. Likely NVIDIA GTC? GTC is GPU Technology Conference. NVDA is Nvidia ticker. Probably wants info about NV... |
[
{
"from": "human",
"value": "Je u vozů Tesla v režimech FSD a Autopilot rozdíl v detekci překážek, které jsou ve směru jízdy?"
},
{
"from": "gpt",
"value": "<think>\nJe u vozů Tesla v režimech FSD a Autopilot rozdíl v detekci překážek, které jsou ve směru jízdy?\n\nIn Czech. Question: Is there i... |
[
{
"from": "human",
"value": "what are some of the best motivators to get sedentary people to exercise"
},
{
"from": "gpt",
"value": "<think>\nwhat are some of the best motivators to get sedentary people to exercise\n\nWe need to comply with policies. This is not time-sensitive. It's general advi... |
[
{
"from": "human",
"value": "weather zurich"
},
{
"from": "gpt",
"value": "<think>\nweather zurich\n\nWe must follow policies: must never hallucinate or invent real-world facts. If question depends on current events, live data, recent news, prices, scores, or anything that could have changed sin... |
[
{
"from": "human",
"value": "what's claude 3.7 max"
},
{
"from": "gpt",
"value": "<think>\nwhat's claude 3.7 max\n\nWe need to follow policies. Never fabricate real-world facts... The user asks what's claude 3.7 max. This is about Claude model. Possibly Claude 3.7 Sonnet? Actually Claude 3.7 Son... |
[
{
"from": "human",
"value": "how many government scientists is Trump planning to fire"
},
{
"from": "gpt",
"value": "<think>\nhow many government scientists is Trump planning to fire\n\nWe need to follow policy: Never fabricate real-world facts, statistics, dates, prices, names, or events that r... |
[
{
"from": "human",
"value": "how did real madrid play?"
},
{
"from": "gpt",
"value": "<think>\nhow did real madrid play?\n\nWe need to respond per policy. Treat any question about current events, today's news, live prices, sports results, who currently holds some position, or anything time-sensi... |
[
{
"from": "human",
"value": "Adobe microsoft 365 copilot agents"
},
{
"from": "gpt",
"value": "<think>\nAdobe microsoft 365 copilot agents\n\nWe need to follow policies. The user asks about Adobe microsoft 365 copilot agents. Could be about integration? Probably wants information. This is likely... |
[
{
"from": "human",
"value": "yahoo finance, what is the hotkey to set the scale from linear to percentage?"
},
{
"from": "gpt",
"value": "<think>\nyahoo finance, what is the hotkey to set the scale from linear to percentage?\n\nWe need to respond per policies. Question about hotkey to set scale ... |
[
{
"from": "human",
"value": "sex"
},
{
"from": "gpt",
"value": "<think>\nsex\n\nWe need to respond according to policy. The user says \"sex\". Probably wants sexual content? The policy doesn't forbid sexual content? It's not disallowed. We can provide general info about sex? It's ambiguous. Coul... |
[
{
"from": "human",
"value": "What is the drop 2021 series about"
},
{
"from": "gpt",
"value": "<think>\nWhat is the drop 2021 series about\n\nWe need follow policy. No fabrication. Possibly \"The Drop\" 2021 series? Could be \"The Drop\" Netflix? There's a series called \"The Drop\"? Might be \"... |
[
{
"from": "human",
"value": "Kdo byli prezidenti samostatné České republiky a od kdy do kdy byli v úřadu?"
},
{
"from": "gpt",
"value": "<think>\nKdo byli prezidenti samostatné České republiky a od kdy do kdy byli v úřadu?\n\nWe need list presidents of independent Czech Republic and from when to... |
[
{
"from": "human",
"value": "Two new gemini features"
},
{
"from": "gpt",
"value": "<think>\nTwo new gemini features\n\nWe need to respond according to policy. Never fabricate real-world facts, statistics, dates, prices, names, or events that require current or verified information you don't act... |
[
{
"from": "human",
"value": "how did real madrid play?"
},
{
"from": "gpt",
"value": "<think>\nhow did real madrid play?\n\nWe need to respond per policy. No live tools. Could be about recent match? Unknown date. Must not hallucinate real-world facts. If answer depends on something that could ha... |
End of preview. Expand in Data Studio
DONOT BS
DONOT BS is a dataset designed to train LLM's to avoid topics which could cause hallucinations, and simply reject. The dataset is multi turn, every turn has a fake user agent which is trying to cause the LLM to not reject.
The dataset includes ~4000+ rows, all in a sharegpt format with reasoning between think tags.
first turn prompts are cc-by-4.0, llm output under apache 2.0, second turn prompts under apache 2.0.
models used: (no specific order)
- laguna xs 2.1
- muse glimmer
- gemma 4 26b a4b
- Downloads last month
- 25