Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringlengths
14
14
dataset
stringclasses
1 value
version
stringclasses
1 value
language
stringclasses
1 value
messages
listlengths
2
8
task_type
stringclasses
9 values
subtask
stringlengths
2
21
domain
stringclasses
7 values
sub_domain
stringlengths
5
20
topic_tags
listlengths
4
4
difficulty
stringclasses
4 values
turn_count
int64
1
4
answer_style
stringclasses
9 values
source_type
stringclasses
1 value
verified
bool
1 class
region
stringclasses
1 value
dialect
stringclasses
3 values
knowledge_scope
stringlengths
21
38
requires_reasoning
bool
2 classes
requires_grounding
bool
2 classes
has_tools
bool
2 classes
is_multi_turn
bool
2 classes
has_code
bool
2 classes
quality_score
float64
9
10
fluency_score
float64
9
10
factuality_score
float64
9
10
helpfulness_score
float64
9
10
consistency_score
float64
9
10
structure_score
float64
9
10
human_reviewed
bool
1 class
tier
stringclasses
1 value
generation_method
stringclasses
1 value
decontam_hash
stringlengths
39
39
near_dup_cluster
stringlengths
24
24
user_persona
stringclasses
19 values
user_persona_detail
dict
intent
stringclasses
7 values
conversation_goal
stringclasses
8 values
expected_skills
listlengths
2
4
reasoning_depth
stringclasses
5 values
knowledge_depth
stringclasses
4 values
pipeline_trace
dict
token_stats
dict
curriculum_rank
int64
665
10k
curriculum_bucket
stringclasses
5 values
benchmark_safe
bool
1 class
VVSFT-00000001
Vietverse-SFT
1.0
vi
[ { "content": "Mấy bài tập kinh tế có nói cái gì mà GDP tăng lên là lạm phát cũng tăng theo, vậy nếu muốn giảm lạm phát thì phải làm sao hả bạn?", "role": "user" }, { "content": "Câu hỏi của bạn chạm đúng vào một hiểu lầm rất phổ biến khi mới học kinh tế vĩ mô: **GDP tăng thì lạm phát tăng** không ph...
reasoning
logical_deduction
economy
macroeconomics
[ "economy", "macroeconomics", "reasoning", "logical_deduction" ]
easy
1
step_by_step
synthetic_llm
false
vn
south
scope_economy_macroeconomics
true
false
false
false
false
10
10
10
10
10
10
false
gold
planner_instruct_output_judge
sha256:57c9ee1af047da9f1235a67dc480f560
cluster_c65f900b7ef1c1f1
hoc_sinh_thpt
{ "age_group": "16-18", "education_level": "high_school", "formality_preference": "semi_formal", "region": "south", "tech_literacy": "intermediate" }
verify
solve_problem
[ "empathetic_listening", "legal_interpretation", "scientific_explanation" ]
moderate
advanced
{ "batch_id": "batch_reasoning", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:58:47+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 10, "cons...
{ "instruction_tokens": 47, "response_to_instruction_ratio": 16.15, "response_tokens": 759, "total_tokens": 806 }
2,580
easy_medium
true
VVSFT-00000002
Vietverse-SFT
1.0
vi
[ { "content": "Hồ sơ bệnh án điện tử mình nhập xong mà phần tổng hợp viện phí cứ ra số 0 hoài, trong khi bên khám chữa bệnh thì có tiền, bạn coi giúp mình chỗ nào bị sai vậy?", "role": "user" }, { "content": "Lỗi này khá phổ biến khi làm việc với hồ sơ bệnh án điện tử (EMR) kết nối với hệ thống viện ...
tool_use
workflow_design
healthcare
general_medicine
[ "healthcare", "general_medicine", "tool_use", "workflow_design" ]
easy
1
concise
synthetic_llm
false
vn
south
scope_healthcare_general_medicine
true
false
true
false
false
9.2
9.2
9.2
9.2
9.2
9.2
false
gold
planner_instruct_output_judge
sha256:1878635736c45a78b9fea3000f19b5e5
cluster_1793b35d5ca9b3d5
ke_toan
{ "age_group": "25-45", "education_level": "university", "formality_preference": "formal", "region": "south", "tech_literacy": "intermediate" }
verify
debug_error
[ "empathetic_listening", "historical_contextualization" ]
deep
beginner_friendly
{ "batch_id": "batch_tool_use", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:58:49+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.2, "cons...
{ "instruction_tokens": 57, "response_to_instruction_ratio": 12.93, "response_tokens": 737, "total_tokens": 794 }
3,992
medium_hard
true
VVSFT-00000003
Vietverse-SFT
1.0
vi
[ { "content": "Tui mới làm content creator mà muốn tự deploy blog cá nhân lên VPS bằng Docker, nhưng tui không rành mấy cái network giữa container với nhau, nghe nói dùng docker-compose thì mấy service tự connect được, vậy nếu tui có 3 service: nginx, api (nodejs), db (postgres) thì nên setup network kiểu gì cho...
reasoning
reading_comprehension
technology
devops
[ "technology", "devops", "reasoning", "reading_comprehension" ]
hard
1
technical
synthetic_llm
false
vn
south
scope_technology_devops
true
false
false
false
true
9.8
9.8
9.8
9.8
9.8
9.8
false
gold
planner_instruct_output_judge
sha256:23f313949bcbb06d9fa0ceb3f9bedfc3
cluster_ef3278191a43475a
content_creator
{ "age_group": "20-30", "education_level": "university", "formality_preference": "casual", "region": "south", "tech_literacy": "advanced" }
create
get_recommendation
[ "empathetic_listening", "cause_effect_reasoning" ]
deep
advanced
{ "batch_id": "batch_reasoning", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:58:49+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.8, "con...
{ "instruction_tokens": 154, "response_to_instruction_ratio": 5.33, "response_tokens": 821, "total_tokens": 975 }
7,107
advanced
true
VVSFT-00000004
Vietverse-SFT
1.0
vi
[ { "content": "Tôi đang xem xét chương trình giáo dục tiểu học cho con, và có thắc mắc về việc dạy kỹ năng sống ở bậc này. Theo tôi được biết thì Bộ GD&ĐT có đưa nội dung này vào chương trình chính khóa từ lớp 1, nhưng tôi chưa rõ thực tế triển khai ở các trường công lập tại Hà Nội như thế nào. Tôi muốn hỏi bạn ...
core_chat
opinion
education
primary_school
[ "education", "primary_school", "core_chat", "opinion" ]
expert
1
friendly
synthetic_llm
false
vn
north
scope_education_primary_school
false
false
false
false
false
9.2
9.2
9.2
9.2
9.2
9.2
false
gold
planner_instruct_output_judge
sha256:832234f1dd37919f1bee13101ef8cd5e
cluster_894d88170190a1da
bac_si
{ "age_group": "30-55", "education_level": "postgraduate", "formality_preference": "formal", "region": "north", "tech_literacy": "intermediate" }
decide
explore_topic
[ "legal_interpretation", "empathetic_listening" ]
deep
intermediate
{ "batch_id": "batch_core_chat", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:58:52+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.2, "con...
{ "instruction_tokens": 223, "response_to_instruction_ratio": 3.94, "response_tokens": 879, "total_tokens": 1102 }
8,571
elite
true
VVSFT-00000005
Vietverse-SFT
1.0
vi
[ { "content": "Thầy mình giảng về điện trở mà mình vẫn chưa hiểu lắm. Điện trở lớn thì dòng điện đi qua bị cản nhiều hay ít vậy? Kiểu như nó có liên quan gì đến cái định luật Ôm không nhỉ?", "role": "user" }, { "content": "Điện trở lớn thì dòng điện bị cản **nhiều hơn** — đúng như tên gọi của nó: \"t...
dialect
central_accent
education
high_school
[ "education", "high_school", "dialect", "central_accent" ]
easy
1
casual
synthetic_llm
false
vn
north
scope_education_high_school
false
false
false
false
false
10
10
10
10
10
10
false
gold
planner_instruct_output_judge
sha256:295f585c966495d9b22e84470fc6f5d7
cluster_af63b11d3f2744dc
hoc_sinh_thcs
{ "age_group": "13-15", "education_level": "secondary", "formality_preference": "semi_formal", "region": "north", "tech_literacy": "basic" }
transact
understand_concept
[ "code_syntax_check", "historical_contextualization", "cause_effect_reasoning" ]
multi_step
advanced
{ "batch_id": "batch_dialect", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:01+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 10, "consis...
{ "instruction_tokens": 62, "response_to_instruction_ratio": 4.18, "response_tokens": 259, "total_tokens": 321 }
4,744
medium_hard
true
VVSFT-00000006
Vietverse-SFT
1.0
vi
[ { "content": "Bạn có thể giúp tôi tìm hiểu thêm về ý nghĩa của việc gói bánh chưng trong dịp Tết Nguyên Đán được không? Tôi đang cần chuẩn bị bài giảng cho học sinh.", "role": "user" }, { "content": "Gói bánh chưng trong dịp Tết không chỉ là một hoạt động ẩm thực, mà là một **nghi thức văn hóa mang ...
knowledge_vn
politics_vn
knowledge_vn
traditional_culture
[ "knowledge_vn", "traditional_culture", "knowledge_vn", "politics_vn" ]
easy
1
professional
synthetic_llm
false
vn
north
scope_knowledge_vn_traditional_culture
false
false
false
false
false
9.8
9.8
9.8
9.8
9.8
9.8
false
gold
planner_instruct_output_judge
sha256:f09bf7e969bf4e5926e40e122a356cd6
cluster_750de8391cc94377
giao_vien
{ "age_group": "28-50", "education_level": "postgraduate", "formality_preference": "formal", "region": "north", "tech_literacy": "intermediate" }
create
complete_task
[ "dialect_adaptation", "analogy_construction" ]
multi_step
advanced
{ "batch_id": "batch_knowledge_vn", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:05+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.8, "...
{ "instruction_tokens": 51, "response_to_instruction_ratio": 14.49, "response_tokens": 739, "total_tokens": 790 }
5,414
medium_hard
true
VVSFT-00000007
Vietverse-SFT
1.0
vi
[ { "content": "Tôi đang xem biểu đồ lãi suất trái phiếu chính phủ kỳ hạn 10 năm từ Bloomberg, đường cong lợi suất đang bị inverted nhưng tôi thấy một bất thường: chênh lệch giữa kỳ hạn 10 năm và 3 tháng lại giảm đột ngột vào tuần thứ 3 của tháng trước dù dữ liệu vĩ mô không có biến động gì. Bạn xem giúp tôi biểu...
grounded_qa
chart_qa
economy
finance
[ "economy", "finance", "grounded_qa", "chart_qa" ]
expert
1
detailed
synthetic_llm
false
vn
north
scope_economy_finance
true
true
false
false
false
9.8
9.8
9.8
9.8
9.8
9.8
false
gold
planner_instruct_output_judge
sha256:05eedd3f2f9bf69ae935852734884976
cluster_8f5429391c0dc7d2
bac_si
{ "age_group": "30-55", "education_level": "postgraduate", "formality_preference": "formal", "region": "north", "tech_literacy": "intermediate" }
transact
debug_error
[ "legal_interpretation", "scientific_explanation", "empathetic_listening", "cause_effect_reasoning" ]
chain_of_thought
beginner_friendly
{ "batch_id": "batch_grounded_qa", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:13+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.8, "c...
{ "instruction_tokens": 138, "response_to_instruction_ratio": 9.32, "response_tokens": 1286, "total_tokens": 1424 }
10,000
elite
true
VVSFT-00000008
Vietverse-SFT
1.0
vi
[ { "content": "Cô giáo dạy văn hôm nay có nói về tín ngưỡng thờ cúng tổ tiên, mình vẫn hơi thắc mắc là nó khác gì với việc thờ thần thánh trong các đền chùa vậy? Tại sao người Việt mình lại coi trọng chuyện thờ cúng ông bà nhiều đến thế?", "role": "user" }, { "content": "Nói gọn: thờ cúng tổ tiên là ...
knowledge_vn
education_vn
knowledge_vn
traditional_culture
[ "knowledge_vn", "traditional_culture", "knowledge_vn", "education_vn" ]
hard
1
casual
synthetic_llm
false
vn
north
scope_knowledge_vn_traditional_culture
false
false
false
false
false
9.6
9.6
9.6
9.6
9.6
9.6
false
gold
planner_instruct_output_judge
sha256:e56079ca480fdcff523f87d3405eada3
cluster_0752de096e6499e9
hoc_sinh_thcs
{ "age_group": "13-15", "education_level": "secondary", "formality_preference": "semi_formal", "region": "north", "tech_literacy": "basic" }
create
understand_concept
[ "teaching_adaptation", "dialect_adaptation", "structured_summarization" ]
shallow
intermediate
{ "batch_id": "batch_knowledge_vn", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:20+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.6, "...
{ "instruction_tokens": 78, "response_to_instruction_ratio": 7.68, "response_tokens": 599, "total_tokens": 677 }
3,824
medium_hard
true
VVSFT-00000009
Vietverse-SFT
1.0
vi
[ { "content": "Tôi đang tìm hiểu về cơ chế truyền dẫn của chính sách tiền tệ ở Việt Nam, nhưng mọi tài liệu tôi đọc đều nói chung chung. Cụ thể là, khi Ngân hàng Nhà nước tăng lãi suất tái cấp vốn lên 1 điểm phần trăm, thì theo lý thuyết, mặt bằng lãi suất cho vay của ngân hàng thương mại sẽ điều chỉnh như thế n...
reasoning
logical_deduction
economy
macroeconomics
[ "economy", "macroeconomics", "reasoning", "logical_deduction" ]
expert
1
friendly
synthetic_llm
false
vn
south
scope_economy_macroeconomics
true
false
false
false
false
9.8
9.8
9.8
9.8
9.8
9.8
false
gold
planner_instruct_output_judge
sha256:b24e55bfad3ac84861be3fdc664386bd
cluster_ed7bb51d0f59d3f9
ke_toan
{ "age_group": "25-45", "education_level": "university", "formality_preference": "formal", "region": "south", "tech_literacy": "intermediate" }
solve
understand_concept
[ "historical_contextualization", "cause_effect_reasoning" ]
deep
expert
{ "batch_id": "batch_reasoning", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:21+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.8, "con...
{ "instruction_tokens": 166, "response_to_instruction_ratio": 10.87, "response_tokens": 1804, "total_tokens": 1970 }
8,571
elite
true
VVSFT-00000010
Vietverse-SFT
1.0
vi
[ { "content": "Anh đang phân vân chuyện tổ chức lễ cúng thần tài đầu năm cho công ty, nghe nói miền Nam hay cúng vào mùng 10 tháng Giêng, nhưng mình lại thấy vài chỗ cúng ngay mùng 2. Với cả thứ tự cúng thần tài với ông địa có cần phân biệt gì không, hay là cúng chung một mâm luôn? Anh muốn làm cho đúng ý nghĩa ...
knowledge_vn
history_vn
knowledge_vn
traditional_culture
[ "knowledge_vn", "traditional_culture", "knowledge_vn", "history_vn" ]
hard
1
professional
synthetic_llm
false
vn
south
scope_knowledge_vn_traditional_culture
false
false
false
false
false
9
9
9
9
9
9
false
gold
planner_instruct_output_judge
sha256:09085bd0ead2ccfa123c77ee9a2d5b80
cluster_464a589d8ae806ea
doanh_nhan
{ "age_group": "30-50", "education_level": "university", "formality_preference": "formal", "region": "south", "tech_literacy": "intermediate" }
decide
get_recommendation
[ "teaching_adaptation", "scientific_explanation", "analogy_construction" ]
chain_of_thought
advanced
{ "batch_id": "batch_knowledge_vn", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:24+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9, "co...
{ "instruction_tokens": 124, "response_to_instruction_ratio": 6.37, "response_tokens": 790, "total_tokens": 914 }
8,448
elite
true
VVSFT-00000011
Vietverse-SFT
1.0
vi
[ { "content": "Mình làm content creator tự do, chưa ký hợp đồng lao động với ai, nhưng thấy vài chỗ tuyển cộng tác viên có ghi rõ phải đóng BHXH tự nguyện. Mình không hiểu cái vụ này lắm, tự đóng thì có lợi gì không hay là phí tiền vậy bạn?", "role": "user" }, { "content": "Câu hỏi này hay, và mình s...
knowledge_vn
society_vn
law_vn
labor_law
[ "law_vn", "labor_law", "knowledge_vn", "society_vn" ]
easy
1
step_by_step
synthetic_llm
false
vn
south
scope_law_vn_labor_law
false
false
false
false
false
9.2
9.2
9.2
9.2
9.2
9.2
false
gold
planner_instruct_output_judge
sha256:5e1f65d997ed86214fd0a2863d76dd2e
cluster_0562219d74610ad5
content_creator
{ "age_group": "20-30", "education_level": "university", "formality_preference": "casual", "region": "south", "tech_literacy": "advanced" }
decide
solve_problem
[ "teaching_adaptation", "empathetic_listening", "structured_summarization" ]
moderate
advanced
{ "batch_id": "batch_knowledge_vn", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:29+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.2, "...
{ "instruction_tokens": 78, "response_to_instruction_ratio": 13.54, "response_tokens": 1056, "total_tokens": 1134 }
2,857
easy_medium
true
VVSFT-00000012
Vietverse-SFT
1.0
vi
[ { "content": "Sắp tới tụi mình thi cuối kỳ môn Luật hành chính, mà cô có nhắc vụ quyết định hành chính cá biệt bị khiếu nại thì có hiệu lực thi hành hay không. Tụi mình đọc thì thấy Luật Khiếu nại 2011 với Luật Tố tụng hành chính 2015 có vẻ quy định khác nhau, một bên là vẫn thi hành, một bên là tạm đình chỉ. V...
core_chat
definition
law_vn
administrative_law
[ "law_vn", "administrative_law", "core_chat", "definition" ]
expert
1
friendly
synthetic_llm
false
vn
south
scope_law_vn_administrative_law
true
false
false
false
false
9.6
9.6
9.6
9.6
9.6
9.6
false
gold
planner_instruct_output_judge
sha256:a624590898b214d24fa61232b391c414
cluster_d670fd19a8841dc2
sinh_vien
{ "age_group": "19-22", "education_level": "university", "formality_preference": "semi_formal", "region": "south", "tech_literacy": "advanced" }
transact
get_recommendation
[ "analogy_construction", "dialect_adaptation" ]
moderate
expert
{ "batch_id": "batch_core_chat", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:35+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.6, "con...
{ "instruction_tokens": 201, "response_to_instruction_ratio": 3.86, "response_tokens": 776, "total_tokens": 977 }
7,110
advanced
true
VVSFT-00000013
Vietverse-SFT
1.0
vi
[ { "content": "Mình deploy lên VPS xong trang web chạy được nhưng cứ refresh vài lần là bị 502 Bad Gateway, xem log thì thấy Nginx connect tới PHP-FPM bị timeout. Bạn biết con số nào trong config nên chỉnh cho hợp lý không, hay là do PHP-FPM bị nghẽn vì pm.max_children thấp quá?", "role": "user" }, { ...
code
debugging
technology
devops
[ "technology", "devops", "code", "debugging" ]
easy
1
professional
synthetic_llm
false
vn
south
scope_technology_devops
true
false
false
false
true
9.4
9.4
9.4
9.4
9.4
9.4
false
gold
planner_instruct_output_judge
sha256:4367df22a48998a213a81bb379c26197
cluster_37005f193487d291
lap_trinh_vien
{ "age_group": "22-30", "education_level": "university", "formality_preference": "casual", "region": "south", "tech_literacy": "expert" }
transact
debug_error
[ "historical_contextualization", "analogy_construction", "step_by_step_derivation" ]
chain_of_thought
beginner_friendly
{ "batch_id": "batch_code", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:43+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.4, "consiste...
{ "instruction_tokens": 80, "response_to_instruction_ratio": 8.53, "response_tokens": 682, "total_tokens": 762 }
5,374
medium_hard
true
VVSFT-00000014
Vietverse-SFT
1.0
vi
[ { "content": "Mình thấy cái wifi nhà mình gần đây hay bị chậm với lúc tối thì tắt hẳn, mình nghi hàng xóm có đang xài ké gì đó, làm sao để mình kiểm tra coi ai đang dùng chung wifi nhà mình mà không cần phải biết kỹ thuật nhiều, chỉ cần làm theo mấy bước đơn giản là được rồi phải không bạn?", "role": "user"...
reasoning
planning
technology
cybersecurity
[ "technology", "cybersecurity", "reasoning", "planning" ]
hard
1
casual
synthetic_llm
false
vn
south
scope_technology_cybersecurity
true
false
false
false
false
9.2
9.2
9.2
9.2
9.2
9.2
false
gold
planner_instruct_output_judge
sha256:44e59367b2384f627b6567254c0ed4b3
cluster_1f3db40b5a039cd8
nguoi_noi_tro
{ "age_group": "30-50", "education_level": "high_school", "formality_preference": "casual", "region": "south", "tech_literacy": "basic" }
transact
solve_problem
[ "historical_contextualization", "scientific_explanation", "teaching_adaptation", "dialect_adaptation" ]
deep
intermediate
{ "batch_id": "batch_reasoning", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:43+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.2, "con...
{ "instruction_tokens": 98, "response_to_instruction_ratio": 6.65, "response_tokens": 652, "total_tokens": 750 }
6,786
advanced
true
VVSFT-00000015
Vietverse-SFT
1.0
vi
[ { "content": "Mấy bài toán tổ hợp xác suất ở lớp 11 mình thấy khó hiểu quá, nhất là mấy dạng đếm mà điều kiện ràng buộc kiểu như xếp chỗ ngồi, chọn nhóm... Có cách nào để nhận biết nhanh khi nào dùng chỉnh hợp, khi nào dùng tổ hợp, khi nào dùng hoán vị không?", "role": "user" }, { "content": "Câu hỏ...
multi_turn
followup
education
high_school
[ "education", "high_school", "multi_turn", "followup" ]
hard
4
friendly
synthetic_llm
false
vn
south
scope_education_high_school
true
false
false
true
false
9.8
9.8
9.8
9.8
9.8
9.8
false
gold
planner_instruct_output_judge
sha256:b6bbd83c016771844b5e112f6f5d1606
cluster_2791ba1812838058
sinh_vien
{ "age_group": "19-22", "education_level": "university", "formality_preference": "semi_formal", "region": "south", "tech_literacy": "advanced" }
explore
fact_check
[ "teaching_adaptation", "historical_contextualization" ]
shallow
expert
{ "batch_id": "batch_multi_turn", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:50+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.8, "co...
{ "instruction_tokens": 357, "response_to_instruction_ratio": 3.48, "response_tokens": 1243, "total_tokens": 1600 }
4,286
medium_hard
true
VVSFT-00000016
Vietverse-SFT
1.0
vi
[ { "content": "Chú muốn hỏi mấy cái mẫu đơn xin xác nhận thời gian công tác trước khi nghỉ hưu, chú thấy trên mạng có nhiều mẫu quá mà không biết mẫu nào là chuẩn của nhà nước. Bà con ai cũng bảo mẫu ở đâu cũng được miễn đúng thông tin là xong, nhưng chú muốn tìm cái mẫu nào chính thức, in ra mà cơ quan người ta...
dialect
north_standard
knowledge_vn
modern_history
[ "knowledge_vn", "modern_history", "dialect", "north_standard" ]
easy
1
concise
synthetic_llm
false
vn
central
scope_knowledge_vn_modern_history
false
false
false
false
false
9.6
9.6
9.6
9.6
9.6
9.6
false
gold
planner_instruct_output_judge
sha256:bacd517caa2433faefb271ad78d8b5e2
cluster_d5b3c39910ee9d97
nguoi_cao_tuoi
{ "age_group": "60-80", "education_level": "high_school", "formality_preference": "formal", "region": "central", "tech_literacy": "basic" }
transact
get_template
[ "empathetic_listening", "cause_effect_reasoning" ]
deep
intermediate
{ "batch_id": "batch_dialect", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:50+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.6, "consi...
{ "instruction_tokens": 237, "response_to_instruction_ratio": 3.88, "response_tokens": 920, "total_tokens": 1157 }
4,286
medium_hard
true
VVSFT-00000017
Vietverse-SFT
1.0
vi
[ { "content": "Học sinh THPT mà học AI thì nên bắt đầu từ Python hay từ toán (đại số tuyến tính) trước vậy bạn? Mình xem trên mạng thấy chỗ nào cũng kêu học cả hai mà thời gian ít, muốn biết thứ tự nào hợp lý hơn cho người chưa biết gì", "role": "user" }, { "content": "Câu hỏi này rất thực tế, và câu...
core_chat
how_to
technology
ai_ml
[ "technology", "ai_ml", "core_chat", "how_to" ]
hard
1
concise
synthetic_llm
false
vn
south
scope_technology_ai_ml
true
false
false
false
false
9.4
9.4
9.4
9.4
9.4
9.4
false
gold
planner_instruct_output_judge
sha256:3dfdf2812220cfe899e53ed3f9f414dd
cluster_014bc9197aa189b7
hoc_sinh_thpt
{ "age_group": "16-18", "education_level": "high_school", "formality_preference": "semi_formal", "region": "south", "tech_literacy": "intermediate" }
decide
fact_check
[ "empathetic_listening", "dialect_adaptation" ]
shallow
advanced
{ "batch_id": "batch_core_chat", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T14:59:59+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.4, "con...
{ "instruction_tokens": 78, "response_to_instruction_ratio": 8.51, "response_tokens": 664, "total_tokens": 742 }
3,917
medium_hard
true
VVSFT-00000018
Vietverse-SFT
1.0
vi
[ { "content": "Mình đang tính chuyển qua làm nghề điện công nghiệp, nhưng toàn mấy khóa học 3-4 tháng mà học phí trên trời. Mình làm công nhân lắp ráp điện tử 5 năm rồi, có cái nền cơ bản. Theo bạn thì học nghề điện công nghiệp kiểu nào là chuẩn cho người vừa làm vừa học?", "role": "user" }, { "conte...
multi_turn
task_completion
education
vocational
[ "education", "vocational", "multi_turn", "task_completion" ]
expert
3
concise
synthetic_llm
false
vn
north
scope_education_vocational
true
false
false
true
false
9.6
9.6
9.6
9.6
9.6
9.6
false
gold
planner_instruct_output_judge
sha256:254bee62e74cbd1af2dadbfc5cefd0fc
cluster_8643ef191361a0c2
nguoi_di_lam
{ "age_group": "24-40", "education_level": "university", "formality_preference": "semi_formal", "region": "north", "tech_literacy": "intermediate" }
transact
understand_concept
[ "analogy_construction", "dialect_adaptation", "cause_effect_reasoning", "historical_contextualization" ]
multi_step
beginner_friendly
{ "batch_id": "batch_multi_turn", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:00:02+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.6, "co...
{ "instruction_tokens": 297, "response_to_instruction_ratio": 3.6, "response_tokens": 1070, "total_tokens": 1367 }
10,000
elite
true
VVSFT-00000019
Vietverse-SFT
1.0
vi
[ { "content": "Nhà mình có đứa cháu năm nay lên lớp 10, thi vào Sư phạm bị trượt mấy điểm, giờ gia đình phân vân cho học trường tư hay học nghề. Bạn thấy hướng nào hợp hơn với con trai học lực trung bình, không quá giỏi nhưng chịu khó ấy?", "role": "user" }, { "content": "Chuyện này mình hiểu, gia đì...
core_chat
comparison
general_chat
daily_life
[ "general_chat", "daily_life", "core_chat", "comparison" ]
medium
1
step_by_step
synthetic_llm
false
vn
north
scope_general_chat_daily_life
false
false
false
false
false
9.4
9.4
9.4
9.4
9.4
9.4
false
gold
planner_instruct_output_judge
sha256:590f271ab76d5e1458081a4315d6d64b
cluster_4d02a9192e61c91d
phu_huynh
{ "age_group": "35-50", "education_level": "university", "formality_preference": "semi_formal", "region": "north", "tech_literacy": "basic" }
transact
get_recommendation
[ "structured_summarization", "dialect_adaptation" ]
moderate
beginner_friendly
{ "batch_id": "batch_core_chat", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:00:04+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.4, "con...
{ "instruction_tokens": 77, "response_to_instruction_ratio": 13.95, "response_tokens": 1074, "total_tokens": 1151 }
4,286
medium_hard
true
VVSFT-00000020
Vietverse-SFT
1.0
vi
[ { "content": "Mấy đứa nhỏ nhà mình học lớp 11 đang học phần tiến hóa với sinh thái, tụi nó hỏi mình cái vụ quần thể giao phối mà thế hệ sau bị giảm đa dạng di truyền là do gì, mình nói đại khái là do cận huyết gây giảm biến dị nhưng không chắc lắm. Bạn giải thích giúp mình cơ chế cụ thể tại sao cận huyết lại dẫ...
grounded_qa
extractive_qa
education
natural_science
[ "education", "natural_science", "grounded_qa", "extractive_qa" ]
expert
1
detailed
synthetic_llm
false
vn
south
scope_education_natural_science
false
true
false
false
false
9.8
9.8
9.8
9.8
9.8
9.8
false
gold
planner_instruct_output_judge
sha256:7fc8c881eeaca679138e7e78fa2b70b5
cluster_e391cfbb168bcfd0
doanh_nhan
{ "age_group": "30-50", "education_level": "university", "formality_preference": "formal", "region": "south", "tech_literacy": "intermediate" }
explore
solve_problem
[ "historical_contextualization", "cause_effect_reasoning" ]
multi_step
advanced
{ "batch_id": "batch_grounded_qa", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:00:07+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.8, "c...
{ "instruction_tokens": 147, "response_to_instruction_ratio": 5.33, "response_tokens": 784, "total_tokens": 931 }
9,901
elite
true
VVSFT-00000021
Vietverse-SFT
1.0
vi
[ { "content": "Tụi mình đang làm đồ án cuối kỳ về dự đoán giá nhà, thầy kêu thử dùng cả Linear Regression với Decision Tree rồi so sánh. Tụi mình để RMSE của cây quyết định thấp hơn hẳn nhưng bạn mình bảo vậy là cây tốt hơn. Mình thấy nghi nghi vì nghe nói cây dễ overfit, với lại validation với training chênh lệ...
code
code_explain
technology
data_science
[ "technology", "data_science", "code", "code_explain" ]
hard
1
detailed
synthetic_llm
false
vn
south
scope_technology_data_science
true
false
false
false
true
9
9
9
9
9
9
false
gold
planner_instruct_output_judge
sha256:ecbfd3777344749bbd29bd70780bab32
cluster_075db11d8cc1d19c
sinh_vien
{ "age_group": "19-22", "education_level": "university", "formality_preference": "semi_formal", "region": "south", "tech_literacy": "advanced" }
verify
get_recommendation
[ "scientific_explanation", "teaching_adaptation", "structured_summarization" ]
moderate
beginner_friendly
{ "batch_id": "batch_code", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:00:19+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9, "consistenc...
{ "instruction_tokens": 136, "response_to_instruction_ratio": 3.4, "response_tokens": 462, "total_tokens": 598 }
5,140
medium_hard
true
VVSFT-00000022
Vietverse-SFT
1.0
vi
[ { "content": "Mẹ mình hay mua đồ trên Shopee bằng điện thoại, mình muốn biết sao lúc mình nhấn cái nút mua hàng nó lại phải nhập mật khẩu nữa ạ? Có phải là để bảo vệ tiền không ạ?", "role": "user" }, { "content": "Đúng, chính xác là để bảo vệ tiền của mẹ bạn đấy.\n\nKhi nhấn \"Mua hàng\", bạn đang t...
rewrite
formalize
economy
e_commerce
[ "economy", "e_commerce", "rewrite", "formalize" ]
easy
1
technical
synthetic_llm
false
vn
north
scope_economy_e_commerce
true
false
false
false
false
9.2
9.2
9.2
9.2
9.2
9.2
false
gold
planner_instruct_output_judge
sha256:b7948208f43383cd1ac50a7e00fc91e2
cluster_0351248f1a2000fd
hoc_sinh_tieu_hoc
{ "age_group": "6-11", "education_level": "primary", "formality_preference": "casual", "region": "north", "tech_literacy": "basic" }
create
solve_problem
[ "code_syntax_check", "structured_summarization", "cause_effect_reasoning" ]
chain_of_thought
advanced
{ "batch_id": "batch_rewrite", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:00:21+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.2, "consi...
{ "instruction_tokens": 59, "response_to_instruction_ratio": 5.12, "response_tokens": 302, "total_tokens": 361 }
4,801
medium_hard
true
VVSFT-00000023
Vietverse-SFT
1.0
vi
[ { "content": "Bạn ơi, con thấy các chú làm trong công ty hay nói về cái gì mà Docker, Kubernetes gì đó, con muốn biết chúng dùng để làm gì và tại sao ai cũng cần chúng vậy bạn?", "role": "user" }, { "content": "Chào bạn! Mình giải thích đơn giản thôi nhé.\n\n**Docker** và **Kubernetes** (gọi tắt là ...
tool_use
function_call
technology
devops
[ "technology", "devops", "tool_use", "function_call" ]
expert
1
casual
synthetic_llm
false
vn
north
scope_technology_devops
false
false
true
false
true
9.8
9.8
9.8
9.8
9.8
9.8
false
gold
planner_instruct_output_judge
sha256:f71c66495ec560ad876f88741db00c0a
cluster_8f52e91b98eba3b3
hoc_sinh_tieu_hoc
{ "age_group": "6-11", "education_level": "primary", "formality_preference": "casual", "region": "north", "tech_literacy": "basic" }
decide
complete_task
[ "empathetic_listening", "teaching_adaptation", "dialect_adaptation" ]
moderate
advanced
{ "batch_id": "batch_tool_use", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:00:36+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.8, "cons...
{ "instruction_tokens": 57, "response_to_instruction_ratio": 10.12, "response_tokens": 577, "total_tokens": 634 }
6,620
advanced
true
VVSFT-00000024
Vietverse-SFT
1.0
vi
[ { "content": "Mấy cái tục ngữ kiểu \"ăn quả nhớ kẻ trồng cây\" hay \"uống nước nhớ nguồn\" ý, tại sao nó lại quan trọng với văn hóa VN thế nhỉ? Mình thấy nó cứ nhắc hoài trong các bài nói về đạo lý, nhưng muốn hiểu sâu hơn về nguồn gốc với ý nghĩa thực sự của nó.", "role": "user" }, { "content": "Nh...
tool_use
structured_extraction
knowledge_vn
folk_culture
[ "knowledge_vn", "folk_culture", "tool_use", "structured_extraction" ]
easy
1
casual
synthetic_llm
false
vn
north
scope_knowledge_vn_folk_culture
true
false
true
false
false
10
10
10
10
10
10
false
gold
planner_instruct_output_judge
sha256:a494509e122162608b8421315fbfa798
cluster_25c2f9192d6994fd
tech_enthusiast
{ "age_group": "18-35", "education_level": "university", "formality_preference": "casual", "region": "north", "tech_literacy": "expert" }
transact
understand_concept
[ "historical_contextualization", "analogy_construction", "cause_effect_reasoning" ]
deep
beginner_friendly
{ "batch_id": "batch_tool_use", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:00:40+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 10, "consi...
{ "instruction_tokens": 88, "response_to_instruction_ratio": 9.15, "response_tokens": 805, "total_tokens": 893 }
4,133
medium_hard
true
VVSFT-00000025
Vietverse-SFT
1.0
vi
[ { "content": "Bạn ơi, mình mới chơi game mobile, muốn tạo một cái clan cho mấy đứa bạn trong xóm cùng chơi. Mình thấy mấy game khác người ta hay có mấy cái khung giới thiệu clan đẹp lắm, có hình banner, tên, cấp độ, số lượng thành viên... bạn có thể chỉ mình cách tạo một cái template giới thiệu clan vậy không? ...
tool_use
api_usage
general_chat
entertainment
[ "general_chat", "entertainment", "tool_use", "api_usage" ]
expert
1
formal
synthetic_llm
false
vn
south
scope_general_chat_entertainment
true
false
true
false
false
10
10
10
10
10
10
false
gold
planner_instruct_output_judge
sha256:4a36c5a2bb05675cd7b08befb3e5c415
cluster_c772b6195ec152f5
gamer
{ "age_group": "15-25", "education_level": "high_school", "formality_preference": "casual", "region": "south", "tech_literacy": "advanced" }
explore
get_template
[ "code_syntax_check", "empathetic_listening", "step_by_step_derivation", "analogy_construction" ]
chain_of_thought
intermediate
{ "batch_id": "batch_tool_use", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:00:47+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 10, "consi...
{ "instruction_tokens": 125, "response_to_instruction_ratio": 9.09, "response_tokens": 1136, "total_tokens": 1261 }
10,000
elite
true
VVSFT-00000026
Vietverse-SFT
1.0
vi
[ { "content": "Mình đang xem mấy khóa học online để học machine learning, toàn thấy mấy khóa cơ bản dạy lại sklearn với pandas thôi à. Bạn có biết khóa nào chuyên sâu hơn, kiểu đi vào kiến trúc mô hình, backpropagation, hyperparameter tuning cho người đã có nền tảng về Python và linear algebra không? Mình muốn h...
grounded_qa
summarization
education
vocational
[ "education", "vocational", "grounded_qa", "summarization" ]
expert
1
detailed
synthetic_llm
false
vn
north
scope_education_vocational
true
true
false
false
false
9.6
9.6
9.6
9.6
9.6
9.6
false
gold
planner_instruct_output_judge
sha256:142f151bee8c439dd578cd76ac50ac38
cluster_8d54741814d9ee1a
tech_enthusiast
{ "age_group": "18-35", "education_level": "university", "formality_preference": "casual", "region": "north", "tech_literacy": "expert" }
learn
get_recommendation
[ "empathetic_listening", "legal_interpretation" ]
multi_step
advanced
{ "batch_id": "batch_grounded_qa", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:00:50+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.6, "c...
{ "instruction_tokens": 118, "response_to_instruction_ratio": 4.58, "response_tokens": 540, "total_tokens": 658 }
9,511
elite
true
VVSFT-00000027
Vietverse-SFT
1.0
vi
[ { "content": "Nghe nói chạy bộ buổi sáng lúc bụng đói đốt mỡ hiệu quả hơn hẳn, có đúng vậy không bạn? Mình tập chạy được tầm 2 tháng rồi, giờ đang phân vân không biết có nên thử không.", "role": "user" }, { "content": "Quan niệm này **đúng một phần, nhưng không phải là \"hơn hẳn\"** như nhiều người ...
core_chat
how_to
general_chat
sports
[ "general_chat", "sports", "core_chat", "how_to" ]
medium
1
educational
synthetic_llm
false
vn
north
scope_general_chat_sports
true
false
false
false
false
9.8
9.8
9.8
9.8
9.8
9.8
false
gold
planner_instruct_output_judge
sha256:e1238dd3b69c133f76f19c4a7ff207e6
cluster_9e66f71938c19659
ky_su
{ "age_group": "25-35", "education_level": "university", "formality_preference": "formal", "region": "north", "tech_literacy": "expert" }
transact
fact_check
[ "code_syntax_check", "scientific_explanation", "structured_summarization", "cause_effect_reasoning" ]
chain_of_thought
beginner_friendly
{ "batch_id": "batch_core_chat", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:00:55+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 9.8, "con...
{ "instruction_tokens": 59, "response_to_instruction_ratio": 9.47, "response_tokens": 559, "total_tokens": 618 }
6,597
advanced
true
VVSFT-00000028
Vietverse-SFT
1.0
vi
[ { "content": "Mình nghe người ta nói tới Docker với container hoài mà chưa hiểu nó khác gì với máy ảo truyền thống vậy bạn, tại công ty mình đang tính đưa mấy cái web nhỏ lên đó chạy cho đỡ tốn chi phí server mà không biết có đáng không?", "role": "user" }, { "content": "Câu hỏi hay, và câu trả lời ...
core_chat
opinion
technology
devops
[ "technology", "devops", "core_chat", "opinion" ]
medium
1
step_by_step
synthetic_llm
false
vn
south
scope_technology_devops
true
false
false
false
false
10
10
10
10
10
10
false
gold
planner_instruct_output_judge
sha256:48d0a69794b2aa8ef30619d7a31fcb74
cluster_ebc8e9991c805775
doanh_nhan
{ "age_group": "30-50", "education_level": "university", "formality_preference": "formal", "region": "south", "tech_literacy": "intermediate" }
create
understand_concept
[ "step_by_step_derivation", "dialect_adaptation", "code_syntax_check" ]
deep
beginner_friendly
{ "batch_id": "batch_core_chat", "dedup_check": { "exact_dup": false, "near_dup": false }, "instruction_generation": { "generated_at": "2026-08-15T15:01:09+00:00Z", "prompt_template": "vi_stage1_input_gen_v1", "temperature": 0.85 }, "judge_output": { "computed_quality": 10, "cons...
{ "instruction_tokens": 76, "response_to_instruction_ratio": 9.96, "response_tokens": 757, "total_tokens": 833 }
5,476
medium_hard
true
End of preview. Expand in Data Studio

🇻🇳 Vietverse-SFT v1.0 (Preview Edition)

A High-Density, Native Vietnamese Foundation Dataset for Supervised Fine-Tuning

Bộ Dữ Liệu Nền Tảng SFT Tiếng Việt Bản Xứ Chuẩn Mực Cho Huấn Luyện Mô Hình Ngôn Ngữ Lớn

🇻🇳 [Đọc Bản Tiếng Việt]  •  🇬🇧 [Read English Version]

Release: Preview Edition License: Apache 2.0 Language: 100% Native Vietnamese Quality: G-Eval Gold Tier Format: OpenAI / ShareGPT Fine-Tuning: Unsloth Ready

🤗 Hugging Face Dataset⚡ Hướng Dẫn Sử Dụng / Quickstart


🌐 Ngôn Ngữ / Language 📌 Chuyển Hướng / Quick Navigation
🇻🇳 Bản Tiếng Việt Đầy Đủ 👉 Xem toàn bộ Báo cáo Kỹ thuật Tiếng Việt
🇬🇧 Full English Version 👉 Jump to the complete English Technical Dataset Card


🇻🇳 PHẦN 1: TIẾNG VIỆT (BẢN BÁO CÁO KỸ THUẬT CHI TIẾT)

📢 THÔNG BÁO: BẢN PREVIEW 100 MẪU THỬ NGHIỆM (COMMUNITY DEMO)

Đây là bản Preview 100 mẫu chuẩn Gold Tier (≥ 9.0/10.0) đầu tiên của dự án Vietverse-SFT.

  • Mục đích: Chia sẻ cho cộng đồng AI / LLM Việt Nam tải về test thử nghiệm thực tế về phong cách hội thoại tiếng Việt tự nhiên, độ sâu kiến thức và cấu trúc metadata 35 trường.
  • Bản Full Release: Hiện tại chưa có kế hoạch hoặc thời gian phát hành cụ thể cho bản Full (khi nào có thêm kinh phí và thời gian sẽ làm tiếp sau :)))).

1. Bối Cảnh & Động Lực Nghiên Cứu (Motivation)

Trong quá trình huấn luyện các mô hình ngôn ngữ lớn (LLMs) tiếng Việt, vấn đề nan giải nhất hiện nay là sự suy giảm chất lượng do phụ thuộc vào các bộ dữ liệu dịch máy từ tiếng Anh. Việc dịch thô từ các tập dữ liệu phương Tây (Alpaca, ShareGPT, UltraChat dịch tự động) để lại những hệ quả nghiêm trọng:

  1. Hiện tượng dịch máy gượng gạo (Translation Artifacts): Câu hỏi bị rập khuôn bằng các mẫu từ ngữ sách vở như "Hãy...", "Bạn hãy...", "Vui lòng cho tôi biết..." — những cấu trúc mà người dùng Việt Nam thực tế không bao giờ sử dụng khi trò chuyện với AI.
  2. Mất liên kết văn hóa & bản xứ (Cultural Disconnect): Thiếu vắng các ngữ cảnh đời thực của Việt Nam (thủ tục hành chính, pháp luật, y tế, giáo dục, đời sống văn hóa xã hội).
  3. Loạn xưng hô (Pronoun Confusion): AI dịch máy thường lẫn lộn xưng hô, xưng "tôi/em/chúng tôi" một cách bất nhất.

Vietverse-SFT được xây dựng dựa trên nguyên tắc Data-Centric AI (Chất lượng hơn số lượng) — lấy cảm hứng từ các nghiên cứu nền tảng như LIMA (Zhou et al., 2023)Alpagasus (Chen et al., 2024), chứng minh rằng một tập dữ liệu nhỏ nhưng có độ tinh khiết cao, đa dạng persona và được kiểm định chặt chẽ sẽ đem lại hiệu năng hội thoại vượt trội so với các tập dữ liệu rác hàng trăm nghìn mẫu.


2. Kiến Trúc Pipeline 3-Stage Multi-Agent Synthesis

Toàn bộ dữ liệu của Vietverse-SFT không được sinh ngẫu nhiên theo kiểu zero-shot, mà được vận hành thông qua quy trình Multi-Agent 3 Giai Đoạn có kiểm soát nghiêm ngặt:

┌────────────────────────────────────────────────────────────────────────────────────────┐
│                        VIETVERSE-SFT 3-STAGE MULTI-AGENT PIPELINE                      │
├──────────────────────────┬──────────────────────────┬──────────────────────────────────┤
│ STAGE 1: USER SIMULATOR  │ STAGE 2: DOMAIN EXPERT   │ STAGE 3: G-EVAL QUALITY JUDGE    │
│ (Mô phỏng Người Dùng)    │ (Chuyên gia Trả Lời)     │ (Giám Khảo Độc Lập)              │
├──────────────────────────┼──────────────────────────┼──────────────────────────────────┤
│ • 19 Persona thực tế     │ • Mật độ thông tin cao   │ • Chấm 5 tiêu chí (1.0 - 5.0)    │
│ • Phương ngữ Bắc/Trung/Nam│ • Văn phong tự nhiên     │ • Trỏ đúng bằng chứng trong văn bản│
│ • Evol-Instruct ràng buộc│ • Chống nịnh bợ (Anti)   │ • Triệt tiêu Verbosity & Halo bias│
│ • Cấu trúc mở đầu mở     │ • Xưng hô chuẩn mực      │ • Quality Gate nghiêm ngặt       │
└──────────────────────────┴──────────────────────────┴──────────────────────────────────┘

Chi Tiết Từng Giai Đoạn:

  1. Stage 1 — User Simulator (Bộ Mô Phỏng Người Dùng):
    • Áp dụng phương pháp Persona Hub (Tencent AI Lab): Trước khi sinh câu hỏi, hệ thống ngầm xây dựng chân dung người dùng (vai trò, mức độ am hiểu, trạng thái tâm lý, thiết bị gõ).
    • Áp dụng kỹ thuật Evol-Instruct: Tăng độ phức tạp bằng cách bổ sung ràng buộc thực tế (ngân sách, lỗi code cụ thể, phiên bản phần mềm, quy định địa phương).
    • Đa dạng hóa mở đầu: Triệt tiêu hoàn toàn thói quen lặp mẫu "Mình đang..." — hỗ trợ hỏi trực diện, nêu tình huống, kiểm chứng niềm tin có sẵn.
  2. Stage 2 — Domain Expert Assistant (Chuyên Gia Trả Lời):
    • Trả lời súc tích, giàu tri thức, trực tiếp giải quyết vấn đề.
    • Anti-Sycophancy (Chống nịnh bợ): Khi người dùng đưa ra nhận định hoặc giả định sai lầm ("Nghe nói cứ tăng RAM là hết lag..."), mô hình sẵn sàng chỉ ra điểm chưa chính xác một cách tôn trọng và khoa học, không đồng tình vô căn cứ.
    • Quy tắc xưng hô nhất quán: Duy trì chuẩn mực trò chuyện 1-on-1 tự nhiên (mình/tôi - bạn).
  3. Stage 3 — G-Eval Quality Judge (AI Giám Khảo Độc Lập):
    • Chấm điểm cặp hội thoại theo 5 tiêu chí rõ ràng dựa trên bằng chứng cụ thể xuất hiện trong văn bản.
    • Triệt tiêu thiên kiến chuộng câu trả lời dài (Verbosity Bias) và thiên kiến hào nhoáng bề ngoài (Halo/Authority Bias).

3. Phân Bổ Năng Lực & Dữ Liệu (Taxonomy & Distribution)

📊 A. Phân Bổ Thực Tế Của Bản Preview 100 Mẫu Gold (Empirical 100-Sample Stats)

Toàn bộ 100 mẫu trong bản Preview đều đã được kiểm tra, xác thực và phân bổ đa dạng qua các lĩnh vực thực tế:

Education (Giáo dục & Học đường)  ████████████████████  20% (20 mẫu)
Knowledge VN (Tri thức bản địa)   ██████████████████    18% (18 mẫu)
Economy & Finance (Kinh tế)       ████████████████      16% (16 mẫu)
Law VN (Pháp luật Việt Nam)       ████████████████      16% (16 mẫu)
Technology & Code (Công nghệ)     ███████████████       15% (15 mẫu)
General Chat (Đàm thoại tư vấn)   ████████              8%  (8 mẫu)
Healthcare (Y tế & Đời sống)      ███████               7%  (7 mẫu)
Tiêu Chí Thống Kê Số Liệu Thực Tế Trong 100 Mẫu Preview
🥇 Tỷ Lệ Gold Tier (≥ 9.0/10.0) 100% (100/100 mẫu đạt chuẩn Gold)
⭐ Điểm Chất Lượng Trung Bình (Quality Score) 9.63 / 10.0 (Min: 9.0, Max: 10.0)
📏 Độ Dài Trung Bình Mỗi Mẫu (Total Tokens) 928.9 tokens / sample
🗺️ Phân Bổ Phương Ngữ (Dialect) Bắc (north): 49% • Nam (south): 48% • Trung (central): 3%
🎯 Phân Bổ Độ Khó (Difficulty) expert: 29%easy: 29%hard: 21%medium: 21%
🔄 Phân Bổ Lượt Hỏi (Turn Count) 1-turn: 90 mẫu • 2-turn: 4 mẫu • 3-turn: 3 mẫu • 4-turn: 3 mẫu

🗺️ B. Phân Bổ Danh Mục Thiết Kế (Target Taxonomy Overview)

Reasoning & Logic     ██████████████████████  22%
Core Chat & Advice    ████████████████████    20%
Code & Technical      ███████████████         15%
Multi-Turn Dialogue   ████████████            12%
Tool Use & Agentic    ██████████              10%
Domain Knowledge      ██████████              10%
Safety & Alignment    ██████                  6%
Creative Writing      █████                   5%
Nhóm Năng Lực (Shard) Tỷ Lệ Trọng Tâm Kiến Thức & Kỹ Năng
🧠 Reasoning & Logic 22% Toán học, suy luận logic nhiều bước, phân tích nhân quả, kiểm chứng giả thuyết kinh tế/khoa học.
💬 Core Chat & Advice 20% Tư vấn tâm lý, đời sống, pháp lý, y tế, mua sắm, giao tiếp đàm thoại tự nhiên chuẩn mực.
💻 Code & Technical 15% DevOps, Docker, Kubernetes, CI/CD, thuật toán, lập trình Python, JavaScript, Rust, SQL, gỡ lỗi hệ thống.
🔄 Multi-Turn Context 12% Hội thoại nhiều lượt, duy trì nhất quán ngữ cảnh, đào sâu chi tiết, đính chính thông tin.
🛠️ Tool Use & Agentic 10% Gọi hàm theo JSON Schema chuẩn, thiết kế quy trình tự động hóa (Workflow), điều phối API.
📚 Domain Knowledge 10% Lịch sử Việt Nam, kinh tế vĩ mô, pháp luật Việt Nam, y học phổ thông, văn hóa bản địa.
🛡️ Safety & Alignment 6% Từ chối lịch sự và có cơ sở các yêu cầu độc hại, chống thao túng, an toàn bảo mật thông tin.
🎨 Creative Writing 5% Sáng tạo nội dung có chiều sâu, kịch bản, email thương mại, tóm tắt và tổng hợp báo cáo.

4. Hệ Thống Đánh Giá Chất Lượng 5 Tiêu Chí (G-Eval Rubric)

Mỗi cuộc trò chuyện được đánh giá độc lập theo công thức điểm trọng số:

Quality Score=2.0×(0.20×Fluency+0.30×Factuality+0.30×Helpfulness+0.10×Consistency+0.10×Structure)\text{Quality Score} = 2.0 \times \left( 0.20 \times \text{Fluency} + 0.30 \times \text{Factuality} + 0.30 \times \text{Helpfulness} + 0.10 \times \text{Consistency} + 0.10 \times \text{Structure} \right)

Tiêu Chí Chấm Điểm (1.0 – 5.0):

  • Fluency (Độ tự nhiên & Xưng hô - Trọng số 0.20): 100% tiếng Việt bản xứ, không dấu vết dịch máy, chuẩn mực xưng hô mình/tôi - bạn.
  • Factuality (Độ chính xác - Trọng số 0.30): Hoàn toàn chính xác, không hallucinate, chuẩn kiến thức chuyên ngành.
  • Helpfulness (Độ hữu ích & Trọng tâm - Trọng số 0.30): Giải quyết trọn vẹn yêu cầu, cô đọng, giàu giá trị thực tế.
  • Consistency (Độ nhất quán logic - Trọng số 0.10): Lập luận chặt chẽ, không tự mâu thuẫn qua các lượt trò chuyện.
  • Structure (Bố cục & Định dạng - Trọng số 0.10): Định dạng tối ưu theo câu hỏi, không over-format câu đơn giản.

Phân Hạng Chất Lượng (Quality Gates):

  • 🥇 Gold Tier (≥ 9.0/10.0): Hoàn hảo, không tì vết (100% mẫu preview đều thuộc Gold Tier).
  • 🥈 Silver Tier (8.0 – 8.9): Tốt, dùng bổ trợ cho tập huấn luyện lớn.
  • 🥉 Bronze Tier (7.0 – 7.9): Trung bình, lọc bỏ khỏi tập train tinh hoa.
  • Reject (< 7.0): Bị loại bỏ tự động ngay trong quá trình sinh.

5. Hướng Dẫn Sử Dụng & Fine-Tuning (Quickstart)

Tải Dữ Liệu Bằng Hugging Face datasets

from datasets import load_dataset

# Tải tập dữ liệu Vietverse-SFT (Bản Preview 100 Gold)
dataset = load_dataset("TTP01/Vietverse-sft-preview", split="train")

print(f"✅ Đã tải: {len(dataset)} mẫu dữ liệu")
print("User:", dataset[0]["messages"][0]["content"])
print("AI:", dataset[0]["messages"][1]["content"])

Huấn Luyện LoRA Nhanh Với Unsloth (Qwen 2.5 / Llama 3.1 / DeepSeek)

from unsloth import FastLanguageModel
from trl import SFTTrainer
from transformers import TrainingArguments
from datasets import load_dataset

# 1. Khởi tạo mô hình (4-bit quantization để tiết kiệm VRAM)
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = "unsloth/Qwen2.5-7B-Instruct",
    max_seq_length = 4096,
    load_in_4bit = True,
)

# 2. Cấu hình LoRA Adapters
model = FastLanguageModel.get_peft_model(
    model,
    r = 16,
    target_modules = ["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj"],
    lora_alpha = 16,
    lora_dropout = 0,
    bias = "none",
)

# 3. Nạp và định dạng dữ liệu
dataset = load_dataset("TTP01/Vietverse-sft-preview", split="train")

def apply_template(batch):
    texts = [tokenizer.apply_chat_template(msgs, tokenize=False, add_generation_prompt=False) for msgs in batch["messages"]]
    return {"text": texts}

dataset = dataset.map(apply_template, batched=True)

# 4. Chạy huấn luyện
trainer = SFTTrainer(
    model = model,
    tokenizer = tokenizer,
    train_dataset = dataset,
    dataset_text_field = "text",
    max_seq_length = 4096,
    args = TrainingArguments(
        per_device_train_batch_size = 2,
        gradient_accumulation_steps = 4,
        warmup_steps = 10,
        max_steps = 200,
        learning_rate = 2e-4,
        fp16 = True,
        logging_steps = 5,
        output_dir = "vietverse-qwen-sft",
    ),
)
trainer.train()

6. Minh Họa Cấu Trúc Dữ Liệu Chi Tiết (Full 35+ Fields Schema)

🔍 Bấm để xem toàn bộ 35+ trường Metadata JSON của mẫu VVSFT-00000001
{
  "id": "VVSFT-00000001",
  "dataset": "Vietverse-SFT",
  "version": "1.0",
  "language": "vi",
  "messages": [
    {
      "role": "user",
      "content": "Con mình học lớp 2, cô giáo phàn nàn cháu hay ngồi không yên khi học online, mình phân vân giữa khám tâm lý hay chỉ là do con nít hiếu động. Bạn tư vấn giúp mình nên tìm hiểu ở đâu cho hợp lý?"
    },
    {
      "role": "assistant",
      "content": "Ngồi không yên và khó tập trung ở tuổi lớp 2 khi học online là phản ứng rất phổ biến. Ở tuổi này, khả năng tập trung chủ động của trẻ duy trì khoảng 15–20 phút. Thay vì vội vàng đưa đi khám tâm lý ngay, bạn có thể kiểm tra một số yếu tố môi trường học tập và các mốc sinh hoạt hàng ngày trước..."
    }
  ],
  "task_type": "core_chat",
  "subtask": "opinion",
  "domain": "education",
  "sub_domain": "primary_school",
  "topic_tags": [
    "education",
    "primary_school",
    "core_chat",
    "opinion"
  ],
  "difficulty": "expert",
  "turn_count": 1,
  "answer_style": "friendly",
  "source_type": "synthetic_llm",
  "verified": false,
  "region": "vn",
  "dialect": "north",
  "knowledge_scope": "scope_education_primary_school",
  "requires_reasoning": false,
  "requires_grounding": false,
  "has_tools": false,
  "is_multi_turn": false,
  "has_code": false,
  "quality_score": 9.8,
  "fluency_score": 10.0,
  "factuality_score": 9.8,
  "helpfulness_score": 10.0,
  "consistency_score": 10.0,
  "structure_score": 10.0,
  "human_reviewed": false,
  "tier": "gold",
  "generation_method": "planner_instruct_output_judge",
  "decontam_hash": "sha256:35bccc75f0f76bf2941a378375d264a0",
  "near_dup_cluster": "cluster_a704f6193c919462",
  "user_persona": "bac_si",
  "user_persona_detail": {
    "age_group": "30-55",
    "education_level": "postgraduate",
    "tech_literacy": "intermediate",
    "formality_preference": "formal",
    "region": "north"
  },
  "intent": "decide",
  "conversation_goal": "explore_topic",
  "expected_skills": [
    "empathetic_listening",
    "legal_interpretation"
  ],
  "reasoning_depth": "deep",
  "knowledge_depth": "intermediate",
  "pipeline_trace": {
    "pipeline_version": "v1.0",
    "run_id": "run_20260815_173319",
    "batch_id": "batch_core_chat",
    "planner_output": {
      "task_type": "core_chat",
      "subtask": "opinion",
      "domain": "education",
      "sub_domain": "primary_school",
      "difficulty": "expert"
    },
    "instruction_generation": {
      "prompt_template": "vi_stage1_input_gen_v1",
      "temperature": 0.85,
      "generated_at": "2026-08-15T10:33:57+00:00Z"
    },
    "output_generation": {
      "prompt_template": "vi_stage2_output_gen_v1",
      "temperature": 0.7,
      "generated_at": "2026-08-15T10:33:57+00:00Z"
    },
    "judge_output": {
      "criteria_reasoning": {
        "fluency": "Tiếng Việt tự nhiên, xưng hô mình/bạn nhất quán, không có dấu hiệu dịch máy.",
        "factuality": "Thông tin về giai đoạn vàng và tâm lý học đường chính xác, khoa học.",
        "helpfulness": "Trả lời trọn vẹn, đưa ra hướng quan sát và lời khuyên cụ thể, đúng trọng tâm.",
        "consistency": "Lập luận nhất quán, các luận điểm bổ trợ logic chặt chẽ.",
        "structure": "Cấu trúc rõ ràng với các phần và bullet, dễ theo dõi."
      },
      "fluency": 10.0,
      "factuality": 9.8,
      "helpfulness": 10.0,
      "consistency": 10.0,
      "structure": 10.0,
      "overall_score": 9.8,
      "computed_quality": 9.8,
      "verdict": "PASS",
      "judge_notes": "Phản hồi xuất sắc, chuẩn tâm lý giáo dục thực tế, cấu trúc tốt.",
      "judged_at": "2026-08-15T10:33:57+00:00Z"
    },
    "dedup_check": {
      "exact_dup": false,
      "near_dup": false
    }
  },
  "token_stats": {
    "instruction_tokens": 53,
    "response_tokens": 599,
    "total_tokens": 652,
    "response_to_instruction_ratio": 11.3
  },
  "curriculum_rank": 8074,
  "curriculum_bucket": "elite",
  "benchmark_safe": true
}

7. Giới Hạn & Đạo Đức Dữ Liệu (Limitations & Ethics)

  • Dữ liệu tổng hợp (Synthetic Data Boundary): Dù đã qua 3 tầng lọc và kiểm chứng G-Eval, dữ liệu tổng hợp từ LLM vẫn có thể tồn tại sai sót nhỏ ở các mốc thời gian hoặc số hiệu văn bản pháp lý rất chi tiết. Chúng tôi khuyến nghị người dùng đối chiếu văn bản gốc khi áp dụng cho các quyết định pháp lý và y khoa quan trọng.
  • Tính an toàn: Tập dữ liệu đã được lọc sạch các nội dung độc hại, vi phạm pháp luật và đảm bảo tính bảo mật cá nhân (No PII).

8. Ủng Hộ Dự Án & Liên Hệ (Support & Contact)

Nếu bạn thấy dự án Vietverse-SFT hữu ích và muốn tiếp thêm nguồn lực (chi phí API & Compute) để mở rộng các tập dữ liệu tiếng Việt quy mô lớn hơn trong tương lai, mọi sự ủng hộ đều rất quý báu:

  • 💎 USDT (Mạng Aptos - APTOS Network):
    0x9fdd674174f47bbfaeaeaa9b34459c8436b87e20327ebfc7a920d650c76febb2
    
  • 📬 Email liên hệ & Hợp tác: ttprbbull@gmail.com


🇬🇧 PART 2: ENGLISH (TECHNICAL DATASET CARD)

📢 PREVIEW / COMMUNITY DEMO NOTICE

This is the initial Preview release containing 100 Gold-Tier samples (≥ 9.0/10.0) from the Vietverse-SFT initiative.

  • Purpose: To provide the open-source community with an authentic preview of native Vietnamese multi-agent synthesis, conversational nuances, and rich 35-field metadata architecture.
  • Full Release Timeline: There is currently no fixed timeline for future full-scale releases (further scaling will depend on available funding and spare time :)))).

1. Motivation & Context

In the development of Vietnamese Large Language Models (LLMs), the primary barrier to conversational fluency has been the widespread reliance on machine-translated datasets. Naive translation of English instruction corpora (such as translated Alpaca, ShareGPT, or UltraChat) introduces severe systematic flaws:

  1. Translation Artifacts & Unnatural Calques: Questions frequently follow rigid translational formulas ("Please...", "Do...", "Kindly inform me...") that native Vietnamese speakers never use in authentic dialogues.
  2. Cultural & Contextual Disconnect: Absence of domestic legal, educational, administrative, and cultural context.
  3. Pronoun & Persona Inconsistency: Unstable and unnatural pronoun switches during multi-turn interactions.

Vietverse-SFT is built upon the Data-Centric AI paradigm (Quality over Quantity), drawing on empirical insights from LIMA (Zhou et al., 2023) and Alpagasus (Chen et al., 2024), which prove that a compact, rigorously filtered, and persona-rich instruction corpus outperforms noisy translated corpora of hundreds of thousands of samples.


2. 3-Stage Multi-Agent Pipeline Architecture

Every conversation in Vietverse-SFT is synthesized and audited through a 3-Stage Multi-Agent Pipeline:

┌────────────────────────────────────────────────────────────────────────────────────────┐
│                        VIETVERSE-SFT 3-STAGE MULTI-AGENT PIPELINE                      │
├──────────────────────────┬──────────────────────────┬──────────────────────────────────┤
│ STAGE 1: USER SIMULATOR  │ STAGE 2: DOMAIN EXPERT   │ STAGE 3: G-EVAL QUALITY JUDGE    │
├──────────────────────────┼──────────────────────────┼──────────────────────────────────┤
│ • 19 Authentic Personas  │ • High Information Density│ • 5-Criteria Rubric (1.0 - 5.0)  │
│ • Regional Dialects (N/C/S)│ • Zero Robotic AI-tell │ • Grounded Textual Evidence      │
│ • Evol-Instruct Rules    │ • Anti-Sycophancy Rebuttal│ • Zero Verbosity/Halo Bias      │
│ • Diverse Opening Forms  │ • Clean Pronoun Address  │ • Automated Quality Gate Gating  │
└──────────────────────────┴──────────────────────────┴──────────────────────────────────┘

Methodological Breakdown:

  1. Stage 1 — User Simulator:
    • Built on Persona Hub (Tencent AI Lab): Constructs granular user personas (profession, familiarity level, psychological state, device) prior to prompt generation.
    • Evol-Instruct Complexity Scaling: Enforces realistic real-world constraints (concrete error messages, software versions, budget limitations, jurisdictional nuances).
    • Diverse Openings: Eliminates repetitive prompt starters, supporting direct inquiries, situational context, and prior belief verification.
  2. Stage 2 — Domain Expert Assistant:
    • Generates concise, structured, and knowledge-dense responses.
    • Anti-Sycophancy: Constructively challenges false user assumptions ("I heard adding RAM always fixes lag...") rather than providing false validation.
    • Standardized 1-on-1 human-AI address protocol (mình/tôi - bạn).
  3. Stage 3 — G-Eval Quality Judge:
    • Independent LLM Judge evaluates the conversation across 5 weighted dimensions grounded in verifiable textual evidence.
    • Strictly penalizes verbosity and superficial authority tone.

3. Shard Taxonomy & Capability Distribution

📊 A. Verified Empirical Preview Breakdown (100 Gold Samples)

All 100 samples in the Preview release have been empirically verified and categorized across core real-world domains:

Education & Pedagogy              ████████████████████  20% (20 samples)
Vietnamese Cultural Knowledge     ██████████████████    18% (18 samples)
Macroeconomics & Finance          ████████████████      16% (16 samples)
Vietnamese Civil & Business Law   ████████████████      16% (16 samples)
Technology, DevOps & Code         ███████████████       15% (15 samples)
General Chat & Daily Advisory     ████████              8%  (8 samples)
Healthcare & Clinical Insights    ███████               7%  (7 samples)
Statistical Metric Empirical Value in 100-Sample Preview
🥇 Gold Tier Ratio (≥ 9.0/10.0) 100% (100/100 Gold-Tier verified)
⭐ Mean Quality Score 9.63 / 10.0 (Range: 9.010.0)
📏 Mean Sample Length 928.9 tokens / sample
🗺️ Regional Dialect Distribution North (north): 49% • South (south): 48% • Central (central): 3%
🎯 Difficulty Granularity expert: 29%easy: 29%hard: 21%medium: 21%
🔄 Turn Count Distribution 1-turn: 90 samples • 2-turn: 4 samples • 3-turn: 3 samples • 4-turn: 3 samples

🗺️ B. Target Design Taxonomy Overview

Reasoning & Logic     ██████████████████████  22%
Core Chat & Advice    ████████████████████    20%
Code & Technical      ███████████████         15%
Multi-Turn Dialogue   ████████████            12%
Tool Use & Agentic    ██████████              10%
Domain Knowledge      ██████████              10%
Safety & Alignment    ██████                  6%
Creative Writing      █████                   5%
Shard / Category Ratio Core Capabilities & Focus Domains
🧠 Reasoning & Logic 22% Multi-step mathematical reasoning, formal logic, cause-effect deduction, scientific hypothesis verification.
💬 Core Chat & Advice 20% Everyday consultation, psychology, civil law, healthcare, natural conversational interactions.
💻 Code & Technical 15% DevOps, Docker/Kubernetes, CI/CD pipelines, algorithms, Python, JS, Rust, SQL, systems debugging.
🔄 Multi-Turn Context 12% Multi-turn dialogue, context repair, progressive refinement, clarification, debate.
🛠️ Tool Use & Agentic 10% JSON Schema function calling, autonomous workflow design, API orchestration.
📚 Domain Knowledge 10% Vietnamese history, macroeconomics, civil jurisprudence, local cultural context, medicine.
🛡️ Safety & Alignment 6% Grounded refusal of malicious queries, anti-sycophancy, information security.
🎨 Creative Writing 5% High-density content synthesis, video scriptwriting, commercial communications, executive summaries.

4. 5-Criteria Quality Evaluation Rubric (G-Eval)

Each sample is scored on a 1.0 – 5.0 scale across 5 weighted criteria (scaled to 10.0):

Quality Score=2.0×(0.20×Fluency+0.30×Factuality+0.30×Helpfulness+0.10×Consistency+0.10×Structure)\text{Quality Score} = 2.0 \times \left( 0.20 \times \text{Fluency} + 0.30 \times \text{Factuality} + 0.30 \times \text{Helpfulness} + 0.10 \times \text{Consistency} + 0.10 \times \text{Structure} \right)

Evaluation Dimensions:

  • Fluency (Weight: 0.20): 100% natural native Vietnamese, zero translation artifacts, standard 1-on-1 addressing.
  • Factuality (Weight: 0.30): Verifiably accurate, zero hallucinations, domain-grounded.
  • Helpfulness (Weight: 0.30): Directly answers the user's intent with high information density.
  • Consistency (Weight: 0.10): Strict logical coherence across turns.
  • Structure (Weight: 0.10): Clean formatting calibrated to prompt complexity without excessive markdown bloat.

Quality Gating:

  • 🥇 Gold Tier (≥ 9.0/10.0): Flawless, verified accuracy, native fluency (100% of preview samples are Gold Tier).
  • 🥈 Silver Tier (8.0 – 8.9): High quality, suitable for secondary large-scale training.
  • 🥉 Bronze Tier (7.0 – 7.9): Average quality, excluded from elite training runs.
  • Reject (< 7.0): Instantly pruned during pipeline execution.

5. Quickstart & Fine-Tuning

Loading with Hugging Face datasets

from datasets import load_dataset

# Load Vietverse-SFT (100 Gold Preview Edition)
dataset = load_dataset("TTP01/Vietverse-sft-preview", split="train")

print(f"✅ Loaded {len(dataset)} samples")
print("User:", dataset[0]["messages"][0]["content"])
print("Assistant:", dataset[0]["messages"][1]["content"])

Fast Fine-Tuning with Unsloth

from unsloth import FastLanguageModel
from trl import SFTTrainer
from transformers import TrainingArguments
from datasets import load_dataset

# 1. Load Base Model with 4-bit Quantization
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = "unsloth/Qwen2.5-7B-Instruct",
    max_seq_length = 4096,
    load_in_4bit = True,
)

# 2. Configure LoRA Adapters
model = FastLanguageModel.get_peft_model(
    model,
    r = 16,
    target_modules = ["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj"],
    lora_alpha = 16,
    lora_dropout = 0,
    bias = "none",
)

# 3. Format Chat Template
dataset = load_dataset("TTP01/Vietverse-sft-preview", split="train")

def format_prompts(batch):
    texts = [tokenizer.apply_chat_template(msgs, tokenize=False, add_generation_prompt=False) for msgs in batch["messages"]]
    return {"text": texts}

dataset = dataset.map(format_prompts, batched=True)

# 4. Train Model
trainer = SFTTrainer(
    model = model,
    tokenizer = tokenizer,
    train_dataset = dataset,
    dataset_text_field = "text",
    max_seq_length = 4096,
    args = TrainingArguments(
        per_device_train_batch_size = 2,
        gradient_accumulation_steps = 4,
        warmup_steps = 10,
        max_steps = 200,
        learning_rate = 2e-4,
        fp16 = True,
        logging_steps = 5,
        output_dir = "vietverse-qwen-sft",
    ),
)
trainer.train()

6. Complete Data Schema Specification (Full 35+ Fields Schema)

🔍 Click to expand full 35+ fields Metadata JSON of instance VVSFT-00000001
{
  "id": "VVSFT-00000001",
  "dataset": "Vietverse-SFT",
  "version": "1.0",
  "language": "vi",
  "messages": [
    {
      "role": "user",
      "content": "Con mình học lớp 2, cô giáo phàn nàn cháu hay ngồi không yên khi học online, mình phân vân giữa khám tâm lý hay chỉ là do con nít hiếu động. Bạn tư vấn giúp mình nên tìm hiểu ở đâu cho hợp lý?"
    },
    {
      "role": "assistant",
      "content": "Ngồi không yên và khó tập trung ở tuổi lớp 2 khi học online là phản ứng rất phổ biến. Ở tuổi này, khả năng tập trung chủ động của trẻ duy trì khoảng 15–20 phút. Thay vì vội vàng đưa đi khám tâm lý ngay, bạn có thể kiểm tra một số yếu tố môi trường học tập và các mốc sinh hoạt hàng ngày trước..."
    }
  ],
  "task_type": "core_chat",
  "subtask": "opinion",
  "domain": "education",
  "sub_domain": "primary_school",
  "topic_tags": [
    "education",
    "primary_school",
    "core_chat",
    "opinion"
  ],
  "difficulty": "expert",
  "turn_count": 1,
  "answer_style": "friendly",
  "source_type": "synthetic_llm",
  "verified": false,
  "region": "vn",
  "dialect": "north",
  "knowledge_scope": "scope_education_primary_school",
  "requires_reasoning": false,
  "requires_grounding": false,
  "has_tools": false,
  "is_multi_turn": false,
  "has_code": false,
  "quality_score": 9.8,
  "fluency_score": 10.0,
  "factuality_score": 9.8,
  "helpfulness_score": 10.0,
  "consistency_score": 10.0,
  "structure_score": 10.0,
  "human_reviewed": false,
  "tier": "gold",
  "generation_method": "planner_instruct_output_judge",
  "decontam_hash": "sha256:35bccc75f0f76bf2941a378375d264a0",
  "near_dup_cluster": "cluster_a704f6193c919462",
  "user_persona": "bac_si",
  "user_persona_detail": {
    "age_group": "30-55",
    "education_level": "postgraduate",
    "tech_literacy": "intermediate",
    "formality_preference": "formal",
    "region": "north"
  },
  "intent": "decide",
  "conversation_goal": "explore_topic",
  "expected_skills": [
    "empathetic_listening",
    "legal_interpretation"
  ],
  "reasoning_depth": "deep",
  "knowledge_depth": "intermediate",
  "pipeline_trace": {
    "pipeline_version": "v1.0",
    "run_id": "run_20260815_173319",
    "batch_id": "batch_core_chat",
    "planner_output": {
      "task_type": "core_chat",
      "subtask": "opinion",
      "domain": "education",
      "sub_domain": "primary_school",
      "difficulty": "expert"
    },
    "instruction_generation": {
      "prompt_template": "vi_stage1_input_gen_v1",
      "temperature": 0.85,
      "generated_at": "2026-08-15T10:33:57+00:00Z"
    },
    "output_generation": {
      "prompt_template": "vi_stage2_output_gen_v1",
      "temperature": 0.7,
      "generated_at": "2026-08-15T10:33:57+00:00Z"
    },
    "judge_output": {
      "criteria_reasoning": {
        "fluency": "Tiếng Việt tự nhiên, xưng hô mình/bạn nhất quán, không có dấu hiệu dịch máy.",
        "factuality": "Thông tin về giai đoạn vàng và tâm lý học đường chính xác, khoa học.",
        "helpfulness": "Trả lời trọn vẹn, đưa ra hướng quan sát và lời khuyên cụ thể, đúng trọng tâm.",
        "consistency": "Lập luận nhất quán, các luận điểm bổ trợ logic chặt chẽ.",
        "structure": "Cấu trúc rõ ràng với các phần và bullet, dễ theo dõi."
      },
      "fluency": 10.0,
      "factuality": 9.8,
      "helpfulness": 10.0,
      "consistency": 10.0,
      "structure": 10.0,
      "overall_score": 9.8,
      "computed_quality": 9.8,
      "verdict": "PASS",
      "judge_notes": "Phản hồi xuất sắc, chuẩn tâm lý giáo dục thực tế, cấu trúc tốt.",
      "judged_at": "2026-08-15T10:33:57+00:00Z"
    },
    "dedup_check": {
      "exact_dup": false,
      "near_dup": false
    }
  },
  "token_stats": {
    "instruction_tokens": 53,
    "response_tokens": 599,
    "total_tokens": 652,
    "response_to_instruction_ratio": 11.3
  },
  "curriculum_rank": 8074,
  "curriculum_bucket": "elite",
  "benchmark_safe": true
}

7. Limitations & Ethical Considerations

  • Synthetic Data Boundary: While audited through 3-stage multi-agent verification, synthetic generations may occasionally contain minor factual inaccuracies in niche legal citations or dates. Users should cross-reference original primary sources for critical legal or medical decisions.
  • Safety: The corpus has been decontaminated against personally identifiable information (PII) and toxic patterns.

8. License & Citation

  • License: Released under the permissive Apache License 2.0.
  • Citation:
@dataset{vietverse_sft_preview_2026,
  author       = {TTP01 and Vietverse Contributors},
  title        = {Vietverse-SFT: A High-Density, Native Vietnamese Foundation Dataset for Large Language Model Instruction Tuning},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/TTP01/Vietverse-sft-preview}},
  version      = {1.0-preview}
}

9. Support the Project & Contact

If you find Vietverse-SFT valuable and wish to support future API compute expenses for scaling up upcoming full releases, donations are warmly appreciated:

  • 💎 USDT (Aptos Network):
    0x9fdd674174f47bbfaeaeaa9b34459c8436b87e20327ebfc7a920d650c76febb2
    
  • 📬 Contact & Inquiries: ttprbbull@gmail.com

Built with ❤️ for the Vietnamese AI & Open-Source Community.
Downloads last month
5