Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringlengths
16
16
task
stringclasses
43 values
source_dataset
stringclasses
9 values
source_lang
stringclasses
23 values
targ_lang
stringclasses
23 values
language
stringclasses
23 values
instruction
stringlengths
0
64.6k
input
stringlengths
0
7.28k
response
stringlengths
1
6.15k
messages
stringlengths
167
66.1k
text
stringlengths
128
65.9k
output_structure
stringlengths
76
999
metadata
stringlengths
64
288
38f7bfbac266cd99
summarization
BeardedMonster/data-gen
eng
twi
eng
Mepa wo kyɛw, m'ani agye sɛ mehu wei. Keeping the Harvest: Preserving Food Without Power Introduction to Food Security Kofi Mensah: Good morning to all our listeners across the Ashanti region. You are tuned into Peace FM Kumasi. Today we are starting a very important conversation about something that affects every si...
Saa krataa no kyerɛa sɛ obi a ɔyɛ agyapadeɛ a ɔdi dwumadum wɔ Ashanti mman mu no betumi agye ne nnuane, titire garden egg, asaa bosome mmua (6 months). Wɔyɛ aduane nnyigye mmuae mmiensa: 1. **Zero Energy Cool Chamber (ZECC)**: Ɛyɛ adan ketewa a wɔde nabo ne nkotoko na ɛyɛ a ɛteɛ sɛ ɛtumi tiri mmua (temperature) koraa ...
[{"role": "user", "content": "Mepa wo kyɛw, m'ani agye sɛ mehu wei. Keeping the Harvest: Preserving Food Without Power\n\nIntroduction to Food Security\n\nKofi Mensah: Good morning to all our listeners across the Ashanti region. You are tuned into Peace FM Kumasi. Today we are starting a very important conversation abo...
<s><|user|>Mepa wo kyɛw, m'ani agye sɛ mehu wei. Keeping the Harvest: Preserving Food Without Power Introduction to Food Security Kofi Mensah: Good morning to all our listeners across the Ashanti region. You are tuned into Peace FM Kumasi. Today we are starting a very important conversation about something that affec...
<|input_lang|><eng><think>The user wants a summary of the provided long document about food preservation in the Ashanti region, specifically focusing on garden eggs. I need to extract the main methods, costs, and the central tension between scientific safety and traditional costs.</think><task_plan><summarize></task_pl...
{"model": "google/gemma-4-31b-it", "domain": "food_cooking", "confidence": 0.95, "tags": ["summarization"], "unsure_src": false, "unsure_tgt": false}
454f9bfdc835356b
summarization
BeardedMonster/data-gen
eng
twi
eng
Me bra, m'ani agye sɛ mehu saa krataa yi. Abebere, m'ani nni hɔ sɛ mɛkenkan nea ɛwɔ mu nyinaa efisɛ m'adwuma pii wɔ hɔ wei. M'asɛ sɛ, The Architecture of Character: A Dialogue on Holistic Child-Rearing Opening Remarks and Introduction Dr. Kofi Mensah: Good morning to everyone gathered here at the community center. On...
Saa krataa yi kyerɛ asɛm a obiجموعة bi ne Dr. Kofi Mensah ne ne mma-ntetee mpanyimfoa bɔɔ ho nkɔmɔ wɔ Ashanti Regional Education Directorate ho. Wɔn asɛm no fa 'Character First' program a wɔyɛa, a ne mfidie ne sɛ wɔbɛma mmofra a wɔwɔ mfe 6 kɔsi 12 mmoa damitue ne suban pa.
[{"role": "user", "content": "Me bra, m'ani agye sɛ mehu saa krataa yi. Abebere, m'ani nni hɔ sɛ mɛkenkan nea ɛwɔ mu nyinaa efisɛ m'adwuma pii wɔ hɔ wei. M'asɛ sɛ, The Architecture of Character: A Dialogue on Holistic Child-Rearing\n\nOpening Remarks and Introduction\n\nDr. Kofi Mensah: Good morning to everyone gathere...
<s><|user|>Me bra, m'ani agye sɛ mehu saa krataa yi. Abebere, m'ani nni hɔ sɛ mɛkenkan nea ɛwɔ mu nyinaa efisɛ m'adwuma pii wɔ hɔ wei. M'asɛ sɛ, The Architecture of Character: A Dialogue on Holistic Child-Rearing Opening Remarks and Introduction Dr. Kofi Mensah: Good morning to everyone gathered here at the community...
<|input_lang|><eng><think>The user wants a summary of the provided long document about the 'Character First' initiative in Akan. I need to capture the core goals, the budget, the mentorship cycle, and the tension regarding textbook funding.</think><task_plan><summarize></task_plan><|target_lang|><aka><response>
{"model": "google/gemma-4-31b-it", "domain": "childhood_parenting", "confidence": 0.95, "tags": ["summarization"], "unsure_src": false, "unsure_tgt": false}
fe8e36ce2b1a6676
financial-analysis
BeardedMonster/data-gen
eng
eng
eng
Got it. Can you remind me to call the choir director tomorrow at 10 AM to discuss the rehearsal schedule?
No wahala, I've set that reminder for you for tomorrow at 10 AM.
[{"role": "system", "content": "You are an AI assistant called Sabi. You have these tools:\n[{\"type\":\"function\",\"function\":{\"name\":\"get_transport_route\",\"description\":\"Find how to travel between two places: route, rough fare, duration.\",\"parameters\":{\"type\":\"object\",\"properties\":{\"origin\":{\"typ...
<s><|system|>You are an AI assistant called Sabi. You have these tools: [{"type":"function","function":{"name":"get_transport_route","description":"Find how to travel between two places: route, rough fare, duration.","parameters":{"type":"object","properties":{"origin":{"type":"string"},"destination":{"type":"string"}}...
<|input_lang|><eng><task_plan><|plan|></task_plan><|target_lang|><eng><response>
{"model": "google/gemma-4-31b-it", "domain": "music", "confidence": 1.0, "tags": ["financial-analysis", "health-education", "summarization", "tool-calling"], "unsure_src": false, "unsure_tgt": false}
a4bb46fb5815f26e
toxicity-spam-detection
BeardedMonster/data-gen
eng
twi
eng
Fair enough. One last thing, can you translate 'Language is the soul of culture' from English into Twi?
Kasamu yɛ amammerɛ nkwa.
[{"role": "user", "content": "I'm researching language revitalization. I have a sentence in Yoruba: 'A n gbiyanju lati tura fun ede wa.' How do I say this in Hausa?"}, {"role": "assistant", "content": "<|input_lang|><yor><task_plan><translate></task_plan><|target_lang|><hau><response>Muna ƙoƙarin raya harshenmu."}, {"r...
<s><|user|>I'm researching language revitalization. I have a sentence in Yoruba: 'A n gbiyanju lati tura fun ede wa.' How do I say this in Hausa?</s><|assistant|><|input_lang|><yor><task_plan><translate></task_plan><|target_lang|><hau><response>Muna ƙoƙarin raya harshenmu.</s><|user|>Thanks. I'm seeing some comments on...
<|input_lang|><eng><task_plan><translate></task_plan><|target_lang|><twi><response>
{"model": "google/gemma-4-31b-it", "domain": "language_linguistics", "confidence": 0.9, "tags": ["toxicity-spam-detection", "translation"], "unsure_src": false, "unsure_tgt": false}
b5a49964f712ad1c
knowledge-boundary
BeardedMonster/data-gen
twi
twi
twi
"Aane, fɛsion noa. Sɛ mepɛ sɛ mehu dɛ m'animdeɛ kɔ so yɛ fɛ, m'adamfo pii kyerɛ me dɛ mɛ(...TRUNCATED)
"Sɛ yɛhwɛ sika a wode tɔa nneɛma no (50, 80, na 120 GHS), ne average yɛ 83.33 GHS. Sɛ saa sik(...TRUNCATED)
"[{\"role\": \"system\", \"content\": \"You are an AI assistant. You have these tools:\\n[{\\\"type\(...TRUNCATED)
"<s><|system|>You are an AI assistant. You have these tools:\n[{\"type\":\"function\",\"function\":{(...TRUNCATED)
<|input_lang|><aka><task_plan><|math|></task_plan><|target_lang|><aka><response>
"{\"model\": \"google/gemma-4-31b-it\", \"domain\": \"youth_life\", \"confidence\": 0.85, \"tags\": (...TRUNCATED)
4e8bde0e2a54acb9
sentiment-analysis
BeardedMonster/data-gen
eng
yor
eng
"This is perfect. I'm sending this to my colleague in Lagos who speaks Yoruba. Can you translate the(...TRUNCATED)
"Àwọn kókó pàtàkì fún SFX tí ó ní owó díẹ̀:\n\n- **Lo àwọn ohun èlò gidi**: Lo(...TRUNCATED)
"[{\"role\": \"user\", \"content\": \"I'm working on a project about Nollywood's evolution. I found (...TRUNCATED)
"<s><|user|>I'm working on a project about Nollywood's evolution. I found this review of a new indie(...TRUNCATED)
<|input_lang|><eng><task_plan><translate></task_plan><|target_lang|><yor><response>
"{\"model\": \"google/gemma-4-31b-it\", \"domain\": \"film_nollywood\", \"confidence\": 0.95, \"tags(...TRUNCATED)
fdcc7b217e21667e
content-writing
BeardedMonster/data-gen
eng
twi
eng
Quick question: what is the specific chemical formula for 'Zirconia-Plus-Ultra-X' dental cement?
"Me nte ase wɔ 'Zirconia-Plus-Ultra-X' ho, na me nni kwan a mebɛhunu ne formula no. Ɛyɛ te sɛ n(...TRUNCATED)
"[{\"role\": \"system\", \"content\": \"You are a helpful AI assistant. You have these tools:\\n[{\\(...TRUNCATED)
"<s><|system|>You are a helpful AI assistant. You have these tools:\n[{\"type\":\"function\",\"funct(...TRUNCATED)
"<|input_lang|><eng><think>The user is asking for a technical chemical formula for a specific produc(...TRUNCATED)
"{\"model\": \"google/gemma-4-31b-it\", \"domain\": \"health_medicine\", \"confidence\": 0.9, \"tags(...TRUNCATED)
21a8401f9716dcc0
insufficient-context
BeardedMonster/data-gen
eng
eng
eng
Fair enough. Can you extract the names of all the cities mentioned in the text I gave you earlier?
The cities mentioned are Accra and Kumasi.
"[{\"role\": \"system\", \"content\": \"You are an AI assistant. You have these tools:\\n[{\\\"type\(...TRUNCATED)
"<s><|system|>You are an AI assistant. You have these tools:\n[{\"type\":\"function\",\"function\":{(...TRUNCATED)
"<|input_lang|><eng><think>The user wants to extract city names from the previously provided text.</(...TRUNCATED)
"{\"model\": \"google/gemma-4-31b-it\", \"domain\": \"tourism_hospitality\", \"confidence\": 1.0, \"(...TRUNCATED)
389336947a7452e0
extractive-qa
BeardedMonster/data-gen
eng
twi
eng
"I'm panicking. Can you find a nearby hospital in Kumasi that has pediatric emergency services? Tell(...TRUNCATED)
"Mehu baaayi mmienu: Komfo Anokye Teaching Hospital (ɛyɛ 2km na wɔka bɔbɔ no anadwo ne awia), n(...TRUNCATED)
"[{\"role\": \"system\", \"content\": \"You are a helpful AI assistant. You have these tools:\\n[{\\(...TRUNCATED)
"<s><|system|>You are a helpful AI assistant. You have these tools:\n[{\"type\":\"function\",\"funct(...TRUNCATED)
"<|input_lang|><eng><think>I found two options. Komfo Anokye is closest. I will list them in Akan.</(...TRUNCATED)
"{\"model\": \"google/gemma-4-31b-it\", \"domain\": \"geography_places\", \"confidence\": 0.85, \"ta(...TRUNCATED)
5ce896416ac7d185
content-writing
BeardedMonster/data-gen
twi
twi
twi
"Abeg, twere m'akyeame bi a mɛka kyerɛ me nnamfo sɛ Monday anɔpa yi, m'aduanepa mpaano no bɛyɛ(...TRUNCATED)
"Sɛnea wopɛ no: \n\n\"Anigye! Monday anɔpa yi, m'aduanepa mpaano no bɛyɛ ahouwo! Yɛatɔ nnoema(...TRUNCATED)
"[{\"role\": \"system\", \"content\": \"You are an AI assistant called Sabi. You have these tools:\\(...TRUNCATED)
"<s><|system|>You are an AI assistant called Sabi. You have these tools:\n[{\"type\":\"function\",\"(...TRUNCATED)
<|input_lang|><aka><task_plan><|generate|><|edit|></task_plan><|target_lang|><aka><response>
"{\"model\": \"google/gemma-4-31b-it\", \"domain\": \"trade_markets\", \"confidence\": 0.92, \"tags\(...TRUNCATED)
End of preview. Expand in Data Studio

No dataset card yet

Downloads last month
23