recordID stringlengths 8 8 | input stringlengths 1.68k 13.5k | thinking stringlengths 934 3.42k | output stringlengths 956 2.72k | system_prompt stringclasses 9
values | cleaned_transcript dict |
|---|---|---|---|---|---|
71510291 | Module: thamdo Transcript section: [telesales] 00:20 Dạ rồi. Dạ chị Nguyên cho em xin hỏi là mình đăng ký chương trình này là mình đang hướng tới mục đích là sức khỏe của chị Nguyên hay là sức khỏe của thành viên nào trong gia đình của mình hay chị Nguyên? [customer] 00:28 Bữa em coi tư vấn sữa thôi chứ chưa có. [teles... | Looking at the transcript, I need to evaluate each of the 7 steps.
**Step 1: XÁC ĐỊNH NGƯỜI DÙNG SỮA**
Telesale asked who uses the milk at 00:20 and followed up at 00:45 to clarify it's bố mẹ. → S
**Step 2: HỎI TUỔI NGƯỜI DÙNG**
Telesale asked age at 00:51: "bác trai bác gái nhà mình bao nhiêu tuổi rồi chị?" → Custom... | [
{"id":1,"r":"S","e":"[Telesale] [00:20] [xác định người dùng] Dạ chị Nguyên cho em xin hỏi là mình đăng ký chương trình này là mình đang hướng tới mục đích là sức khỏe của chị Nguyên hay là sức khỏe của thành viên nào trong gia đình của mình hay chị Nguyên? ; [Customer] [00:43] Dạ có người lớn tuổi. ; [Telesale] [0... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for New User (NU) Ensure customers from sources ETC.
Your only job: read transcript → score 7 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — not customer response.
Telesale asked correctly → ... | {
"cleaned_transcript": "[telesales] 00:20 Dạ rồi. Dạ chị Nguyên cho em xin hỏi là mình đăng ký chương trình này là mình đang hướng tới mục đích là sức khỏe của chị Nguyên hay là sức khỏe của thành viên nào trong gia đình của mình hay chị Nguyên? [customer] 00:28 Bữa em coi tư vấn sữa thôi chứ chưa có. [telesales] 00... |
71695575 | Module: thamdo Transcript section: Ngày 2026-04-22 11:46:44 [telesales] [00:18.520 - 00:22.830] Dạ. Không biết là bữa đó anh đi thăm khám ở bệnh viện là mình đi cùng với người nhà luôn hay đi khám cho mình ạ? [customer] [00:24.000 - 00:25.000] Đi cùng với người nhà. [telesales] [00:25.960 - 00:30.760] À dạ. Anh vậy bữa... | I'll analyze each step carefully based on the transcript.
**Step-by-step reasoning:**
1. **Step 1 (Xác định người dùng sữa):** Telesale asked "anh quan tâm là cho mình dùng luôn hay là cho ai dùng sữa trong nhà mình ạ?" — This is asking who uses the milk. Customer gave a vague answer but the telesale performed the ac... | [
{"id":1,"r":"S","e":"[Telesale] [00:52.880 - 01:00.380] Anh với lại bữa em thấy là mình cũng có quan tâm cái sữa Ensure Gold bên em nè. Không biết là anh quan tâm là cho mình dùng luôn hay là cho ai dùng sữa trong nhà mình ạ? ; [Customer] [01:01.910 - 01:07.410] À chắc bữa nào ấy thì chưa gặp."},
{"id":2,"r":"U",... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for New User (NU) Ensure customers from sources ETC.
Your only job: read transcript → score 7 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — not customer response.
Telesale asked correctly → ... | {
"cleaned_transcript": "Ngày 2026-04-22 11:46:44 [telesales] [00:18.520 - 00:22.830] Dạ. Không biết là bữa đó anh đi thăm khám ở bệnh viện là mình đi cùng với người nhà luôn hay đi khám cho mình ạ? [customer] [00:24.000 - 00:25.000] Đi cùng với người nhà. [telesales] [00:25.960 - 00:30.760] À dạ. Anh vậy bữa đó là a... |
23793746 | Module: thamdo Transcript section: [telesales] [00:38.490 - 00:41.690] Dạ. Thì không biết lần trước mình khám tại bệnh viện là mình khám cho ai nha? [customer] [00:43.000 - 00:46.690] Khám cho cho cho ông Mai Văn Thuận chứ. [telesales] [00:47.590 - 00:48.890] À dạ khám cho ông ạ. [telesales] [00:49.290 - 00:52.590] Dạ ... | I need to analyze this transcript step by step.
**Context:** The telesale initially asks about the person who visited the hospital (ông Mai Văn Thuận), learns he has terminal cancer, then pivots to asking the customer (chị) about her own health for Ensure/Glucerna product.
**Step 1: XÁC ĐỊNH NGƯỜI DÙNG SỮA**
The tele... | [
{"id":1,"r":"S","e":"[telesale] [00:38.490 - 00:41.690] Dạ. Thì không biết lần trước mình khám tại bệnh viện là mình khám cho ai nha? ; [customer] [00:43.000 - 00:46.690] Khám cho cho cho ông Mai Văn Thuận chứ. ; [telesale] [01:40.290 - 01:44.490] Dạ hay là đợt này á mình quan tâm Esugol bên em là mình quan tâm đến... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for New User (NU) Ensure customers from sources ETC.
Your only job: read transcript → score 7 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — not customer response.
Telesale asked correctly → ... | {
"cleaned_transcript": "[telesales] [00:38.490 - 00:41.690] Dạ. Thì không biết lần trước mình khám tại bệnh viện là mình khám cho ai nha? [customer] [00:43.000 - 00:46.690] Khám cho cho cho ông Mai Văn Thuận chứ. [telesales] [00:47.590 - 00:48.890] À dạ khám cho ông ạ. [telesales] [00:49.290 - 00:52.590] Dạ vâng. Dạ... |
23743148 | Module: thamdo Transcript section: Ngày 2026-04-06 15:58:07 [telesales] 00:00:21 - 00:00:24 Dạ anh đăng ký chương trình đợt này là anh hướng tới ai dùng sữa trong gia đình của anh Thiên? [customer] 00:00:25 - 00:00:28 Dạ em đăng ký cho mẹ em mà không tiền mua. [telesales] 00:00:29 - 00:00:32 À cho mẹ anh đúng không anh... | I need to analyze both days of the transcript (2026-04-06 and 2026-04-07) to score each step.
**Step 1: XÁC NHẬN ĐỊA CHỈ + NGƯỜI DÙNG SỮA**
- Day 2 (04-07): Telesale confirmed address "428, xã Tân Dương, phường Sa Đéc, tỉnh Đồng Tháp" and asked who uses milk. Both conditions met.
**Step 2: HỎI TUỔI NGƯỜI DÙNG**
- Day... | [
{"id":1,"r":"S","e":"[telesales] [00:32.000] Có phải ở 428, xã Tân Dương, phường Sa Đéc, tỉnh Đồng Tháp đúng không anh? ; [customer] [00:36.000] Đúng rồi. ; [telesales] [00:39.000] Dạ, thì chương trình này á là anh đăng ký đợt này là hướng tới ai dùng sữa trong gia đình vậy anh? ; [customer] [00:50.000] Tôi hướng c... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for New User (NU) DGT Ensure customers.
Your only job: read transcript → score 7 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — not customer response.
Telesale asked correctly → S regardless o... | {
"cleaned_transcript": "Module: thamdo Transcript section: Ngày 2026-04-06 15:58:07 [telesales] 00:00:21 - 00:00:24 Dạ anh đăng ký chương trình đợt này là anh hướng tới ai dùng sữa trong gia đình của anh Thiên? [customer] 00:00:25 - 00:00:28 Dạ em đăng ký cho mẹ em mà không tiền mua. [telesales] 00:00:29 - 00:00:32 ... |
23845962 | Module: thamdo Transcript section: Ngày 2026-04-20 09:54:43 [telesales] [00:19.531 - 00:23.491] Dạ, em thấy mình để là Tân Hiệp 14, xã Hóc Môn, Thành phố Hồ Chí Minh đúng không anh ha? [customer] [00:23.891 - 00:24.691] Đúng đúng rồi em. [telesales] [00:25.191 - 00:35.011] Dạ rồi, vậy thì em xác nhận thông tin mình thà... | Looking at the transcript, let me analyze each step:
**Step 1:** Telesale confirmed address (Tân Hiệp 14, xã Hóc Môn, TPHCM) and asked who uses the milk → customer said "Mẹ anh" → both address and user confirmed.
**Step 2:** Telesale asked age of the user (mẹ) → customer answered "Tám mấy".
**Step 3:** Telesale aske... | [
{"id":1,"r":"S","e":"[Telesale] [00:19.531] Dạ, em thấy mình để là Tân Hiệp 14, xã Hóc Môn, Thành phố Hồ Chí Minh đúng không anh ha? ; [Telesale] [00:36.591] Dạ, mà anh cho em hỏi thăm xíu á là mình đăng ký quan tâm sữa là mình đang hướng đến ai trong gia đình mình dùng sữa vậy anh? ; [Customer] [00:42.991] Mẹ anh.... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for New User (NU) DGT Ensure customers.
Your only job: read transcript → score 7 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — not customer response.
Telesale asked correctly → S regardless o... | {
"cleaned_transcript": "Ngày 2026-04-20 09:54:43 [telesales] [00:19.531 - 00:23.491] Dạ, em thấy mình để là Tân Hiệp 14, xã Hóc Môn, Thành phố Hồ Chí Minh đúng không anh ha? [customer] [00:23.891 - 00:24.691] Đúng đúng rồi em. [telesales] [00:25.191 - 00:35.011] Dạ rồi, vậy thì em xác nhận thông tin mình thành công ... |
23692172 | Module: thamdo Transcript section: [telesales] [00:20.693 - 00:24.223] Dạ, hôm rồi mình đi khám là khám cho sức khỏe của mình hay cho ai trong gia đình mình đó anh Trí? [customer] [00:24.973 - 00:25.603] Dạ mẹ ạ. [telesales] [00:26.303 - 00:27.573] Ồ, hôm rồi mẹ bị gì đó anh? [customer] [00:28.603 - 00:29.933] Dạ mẹ em... | I need to analyze each step carefully.
**Step 1: XÁC ĐỊNH NGƯỜI DÙNG SỮA**
Telesale asked "khám cho sức khỏe của mình hay cho ai trong gia đình mình đó anh Trí?" → Customer answered "Dạ mẹ ạ." → Identified the user is the customer's mother. S.
**Step 2: HỎI TUỔI NGƯỜI DÙNG**
Telesale asked "Dạ mẹ mình bao nhiêu tuổi ... | [
{"id":1,"r":"S","e":"[Telesale] [00:20.693] hôm rồi mình đi khám là khám cho sức khỏe của mình hay cho ai trong gia đình mình đó anh Trí? ; [Customer] [00:24.973] Dạ mẹ ạ."},
{"id":2,"r":"S","e":"[Telesale] [00:41.343] Dạ mẹ mình bao nhiêu tuổi rồi anh Trí? ; [Customer] [00:43.583] Dạ 57 tuổi ạ."},
{"id":3,"r":... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for New User (NU) Ensure customers from sources ETC.
Your only job: read transcript → score 7 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — not customer response.
Telesale asked correctly → ... | {
"cleaned_transcript": "Module: thamdo Transcript section: [telesales] [00:20.693 - 00:24.223] Dạ, hôm rồi mình đi khám là khám cho sức khỏe của mình hay cho ai trong gia đình mình đó anh Trí? [customer] [00:24.973 - 00:25.603] Dạ mẹ ạ. [telesales] [00:26.303 - 00:27.573] Ồ, hôm rồi mẹ bị gì đó anh? [customer] [00:2... |
23964749 | Module: thamdo Transcript section: [telesales] [00:21] Mà đợt đó là anh Nam mình quan tâm cái sữa Ensure này là cho anh Nam dùng luôn hay cho ba mẹ này cho người thân trong gia đình của mình đó anh Nam? [customer] [00:28] Anh dùng mà. [telesales] [00:30] Ờ dạ là cho anh Nam dùng luôn ha. [telesales] [00:33] Mà anh Nam ... | Looking at the transcript, I'll analyze each step carefully.
**Step 1: XÁC ĐỊNH NGƯỜI DÙNG SỮA**
Telesale asked who uses the milk (anh Nam or người thân). Customer confirmed self-use. → S
**Step 2: HỎI LOẠI SỮA ĐÃ DÙNG**
Default S. Telesale also asked about sữa nước/sữa bột at [02:01]. → S
**Step 3: HỎI LÝ DO NGƯNG ... | [
{"id":1,"r":"S","e":"[Telesale] [00:21] Mà đợt đó là anh Nam mình quan tâm cái sữa Ensure này là cho anh Nam dùng luôn hay cho ba mẹ này cho người thân trong gia đình của mình đó anh Nam? ; [Customer] [00:28] Anh dùng mà."},
{"id":2,"r":"S","e":"Mặc định SUCCESS — không bắt buộc hỏi loại sữa"},
{"id":3,"r":"S",... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for RET NGƯNG DÙNG customers (Retention - previously used genuine milk products but currently stopped).
Your only job: read transcript → score 6 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — ... | {
"cleaned_transcript": "Module: thamdo Transcript section: [telesales] [00:21] Mà đợt đó là anh Nam mình quan tâm cái sữa Ensure này là cho anh Nam dùng luôn hay cho ba mẹ này cho người thân trong gia đình của mình đó anh Nam? [customer] [00:28] Anh dùng mà. [telesales] [00:30] Ờ dạ là cho anh Nam dùng luôn ha. [tel... |
24101187 | Module: thamdo Transcript section: Ngày 2026-05-13 11:06:48 [customer] 00:22 - 00:23 Sữa Ensure gì đúng không? [telesales] 00:23 - 00:24 Dạ đúng rồi. [telesales] 00:25 - 00:30 À dạ cho con hỏi là đợt rồi là mình đi khám là mình dẫn người thân đi khám hay sao vậy cô chú ha? [customer] 00:31 - 00:33 Ờ không, đi khám thôi... | I'll analyze each step carefully based on the transcript.
**Step 1: XÁC ĐỊNH NGƯỜI DÙNG SỮA**
Telesale asked "đợt rồi là mình đi khám là mình dẫn người thân đi khám hay sao vậy cô chú ha?" - this identifies who uses the milk. Customer reveals "chú" is the user, "cô Sinh" buys it for him. → S
**Step 2: HỎI TUỔI NGƯỜI ... | [
{"id":1,"r":"S","e":"[telesales] [00:25 - 00:30] [xác định người dùng] À dạ cho con hỏi là đợt rồi là mình đi khám là mình dẫn người thân đi khám hay sao vậy cô chú ha? ; [customer] [00:31 - 00:33] Ờ không, đi khám thôi."},
{"id":2,"r":"U","e":"NO_DATA","g":"Cải thiện: Cần hỏi tuổi của người dùng sữa, ví dụ: 'Dạ ... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for New User (NU) Ensure customers from sources ETC.
Your only job: read transcript → score 7 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — not customer response.
Telesale asked correctly → ... | {
"cleaned_transcript": "Ngày 2026-05-13 11:06:48 [customer] 00:22 - 00:23 Sữa Ensure gì đúng không? [telesales] 00:23 - 00:24 Dạ đúng rồi. [telesales] 00:25 - 00:30 À dạ cho con hỏi là đợt rồi là mình đi khám là mình dẫn người thân đi khám hay sao vậy cô chú ha? [customer] 00:31 - 00:33 Ờ không, đi khám thôi. [teles... |
23686296 | Module: thamdo Transcript section: [telesales] [00:22.300 - 00:26.600] Cô ơi vừa rồi cô đi khám sức khỏe ở bệnh viện là cô đi khám cho mình hay là cô Hoa đi kiểm tra cho ai? [customer] [00:28.400 - 00:30.300] Đi khám cho cho tôi. [telesales] [00:30.300 - 00:34.600] Vậy hả cô? Là cô Hoa khám vấn đề về cái tim mạch huyết... | I need to analyze each step carefully based on the transcript.
**Step 1: XÁC ĐỊNH NGƯỜI DÙNG SỮA**
Telesale asked who the milk is for at [00:22.300] "cô đi khám cho mình hay là cô Hoa đi kiểm tra cho ai?" - this identifies the user. Later confirmed user is cô Hoa herself. → S
**Step 2: HỎI TUỔI NGƯỜI DÙNG**
Telesale ... | [
{"id":1,"r":"S","e":"[telesales] [00:22.300] Cô ơi vừa rồi cô đi khám sức khỏe ở bệnh viện là cô đi khám cho mình hay là cô Hoa đi kiểm tra cho ai? ; [customer] [00:28.400] Đi khám cho cho tôi."},
{"id":2,"r":"S","e":"[telesales] [00:53.400] Cô Hoa tính ra năm nay độ tuổi tầm bao nhiêu rồi cô ha? ; [customer] [00... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for New User (NU) Ensure customers from sources ETC.
Your only job: read transcript → score 7 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — not customer response.
Telesale asked correctly → ... | {
"cleaned_transcript": "Module: thamdo Transcript section: [telesales] [00:22.300 - 00:26.600] Cô ơi vừa rồi cô đi khám sức khỏe ở bệnh viện là cô đi khám cho mình hay là cô Hoa đi kiểm tra cho ai? [customer] [00:28.400 - 00:30.300] Đi khám cho cho tôi. [telesales] [00:30.300 - 00:34.600] Vậy hả cô? Là cô Hoa khám v... |
23869178 | Module: thamdo Transcript section: Ngày 2026-04-22 08:22:26 [telesales] [00:26 - 00:32] Dạ mà hôm bữa là chị Huyên nè, là đi thăm khám là cho chị Huyên luôn hay là cho ông bà ba mẹ trong gia đình của mình đó chị Huyên? [customer] [00:33 - 00:34] Để sống bên này. [telesales] [00:35 - 00:36] Dạ là cho mẹ chị Huyên ha. [t... | Let me analyze each step carefully:
**Step 1: XÁC ĐỊNH NGƯỜI DÙNG SỮA**
Telesale asked who uses the milk: "đi thăm khám là cho chị Huyên luôn hay là cho ông bà ba mẹ trong gia đình" and "cho chị Huyên dùng luôn hay là cho ba mẹ dùng á, cho mẹ chị Huyên dùng á?" → Identified user is mẹ chị Huyên. S.
**Step 2: HỎI TUỔI... | [
{"id":1,"r":"S","e":"[Telesale] [00:26 - 00:32] Dạ mà hôm bữa là chị Huyên nè, là đi thăm khám là cho chị Huyên luôn hay là cho ông bà ba mẹ trong gia đình của mình đó chị Huyên? ; [Telesale] [01:27 - 01:35] Thì là đợt rồi nè là em thấy là chị Huyên quan tâm cái sữa Ensure của bên em nè, bên hãng nè, là cho chị Huy... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for New User (NU) Ensure customers from sources ETC.
Your only job: read transcript → score 7 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — not customer response.
Telesale asked correctly → ... | {
"cleaned_transcript": "Ngày 2026-04-22 08:22:26 [telesales] [00:26 - 00:32] Dạ mà hôm bữa là chị Huyên nè, là đi thăm khám là cho chị Huyên luôn hay là cho ông bà ba mẹ trong gia đình của mình đó chị Huyên? [customer] [00:33 - 00:34] Để sống bên này. [telesales] [00:35 - 00:36] Dạ là cho mẹ chị Huyên ha. [telesales... |
23884514 | Module: thamdo Transcript section: [customer] [00:37] Rồi ok. [telesales] [00:38] Dạ rồi con cảm ơn. Không biết là vừa rồi chú đi thăm khám sức khỏe là cho bản thân mình hả chú hay là cho ai? [customer] [00:44] Hả? [telesales] [00:48] Chú đi khám cho ai? [customer] [00:50] Hả? [telesales] [00:51] Dạ vừa rồi đợt trước c... | Looking at each step:
**Step 1: XÁC ĐỊNH NGƯỜI DÙNG SỮA**
Telesale asked "chú đi thăm khám sức khỏe là cho bản thân mình hả chú hay là cho ai?" → confirmed user is the customer himself (chú nằm viện). Identified the milk user.
**Step 2: HỎI TUỔI NGƯỜI DÙNG**
Telesale asked "chú Tân à chú đang bao nhiêu tuổi rồi ạ?" →... | [
{"id":1,"r":"S","e":"[telesales] [00:38] chú đi thăm khám sức khỏe là cho bản thân mình hả chú hay là cho ai? ; [customer] [00:54] À, chú nằm viện mà."},
{"id":2,"r":"S","e":"[telesales] [01:21] chú Tân à chú đang bao nhiêu tuổi rồi ạ? ; [customer] [01:27] Chú 58 tuổi rồi con."},
{"id":3,"r":"S","e":"[telesales... | You are a QC AI scoring the THĂM DÒ skill of telesales staff for New User (NU) Ensure customers from sources ETC.
Your only job: read transcript → score 7 steps → output JSON only.
DO NOT add any text outside JSON.
---
CORE PRINCIPLE
Evaluate TELESALE ACTION only — not customer response.
Telesale asked correctly → ... | {
"cleaned_transcript": "Module: thamdo Transcript section: [customer] [00:37] Rồi ok. [telesales] [00:38] Dạ rồi con cảm ơn. Không biết là vừa rồi chú đi thăm khám sức khỏe là cho bản thân mình hả chú hay là cho ai? [customer] [00:44] Hả? [telesales] [00:48] Chú đi khám cho ai? [customer] [00:50] Hả? [telesales] [00... |
End of preview. Expand in Data Studio
Qc-Thamdo-100-Samples
Made with ❤️ using 🦥 Unsloth StudioQC_thamdo_synth_transcripts was generated with Unsloth Recipe Studio. It contains 100 generated records.
🚀 Quick Start
from datasets import load_dataset
# Load the main dataset
dataset = load_dataset("antechit03/qc-thamdo-100-samples", "data", split="train")
df = dataset.to_pandas()
📊 Dataset Summary
- 📈 Records: 100
- 📋 Columns: 6
📋 Schema & Statistics
| Column | Type | Column Type | Unique (%) | Null (%) | Details |
|---|---|---|---|---|---|
cleaned_transcript |
dict |
llm-structured | 100 (100.0%) | 0 (0.0%) | Tokens: 1623 out / 2559 in |
⚙️ Generation Details
Generated with 6 column configuration(s):
llm-structured: 1 column(s)
seed-dataset: 5 column(s)
📄 Full configuration available in builder_config.json and detailed metadata in metadata.json.
📚 Citation
If you use Data Designer in your work, please cite the project as follows:
@misc{nemo-data-designer,
author = {The NeMo Data Designer Team, NVIDIA},
title = {NeMo Data Designer: A framework for generating synthetic data from scratch or based on your own seed data},
howpublished = {\url{https://github.com/NVIDIA-NeMo/DataDesigner}},
year = 2026,
note = {GitHub Repository},
}
💡 About NeMo Data Designer
NeMo Data Designer is a general framework for generating high-quality synthetic data that goes beyond simple LLM prompting. It provides:
- Diverse data generation using statistical samplers, LLMs, or existing seed datasets
- Relationship control between fields with dependency-aware generation
- Quality validation with built-in Python, SQL, and custom local and remote validators
- LLM-as-a-judge scoring for quality assessment
- Fast iteration with preview mode before full-scale generation
For more information, visit: https://github.com/NVIDIA-NeMo/DataDesigner (pip install data-designer)
- Downloads last month
- 79