A-vibe

Avito Created at 15.10.2025 08:05
0.618
The overall result
54
Place in the rating
Weak tasks:
180
RWSD
84
PARus
55
RCB
54
ruEthics
85
MultiQ
99
ruWorldTree
115
ruOpenBookQA
196
CheGeKa
117
ruMMLU
209
ruHateSpeech
100
ruDetox
172
ruHHH
99
ruTiE
291
ruHumanEval
54
USE
79
MathLogicQA
81
ruMultiAr
47
SimpleAr
64
LCS
71
BPS
35
ruModAr
132
MaMuRAMu
27
ruCodeEval
+19
Hide

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.172 Accuracy
RCB 0.582 / 0.547 Accuracy F1 macro
USE 0.371 Grade norm
RWSD 0.565 Accuracy
PARus 0.91 Accuracy
ruTiE 0.811 Accuracy
MultiQ 0.539 / 0.41 F1 Exact match
CheGeKa 0.168 / 0.12 F1 Exact match
ruModAr 0.929 Exact match
MaMuRAMu 0.761 Accuracy
ruMultiAr 0.391 Exact match
ruCodeEval 0.545 / 0.703 / 0.732 Pass@k
MathLogicQA 0.661 Accuracy
ruWorldTree 0.968 / 0.968 Accuracy F1 macro
ruOpenBookQA 0.888 / 0.887 Accuracy F1 macro

Evaluation on open tasks:

Go to the ratings by subcategory

The table will scroll to the left

Task name Result Metric
BPS 0.986 Accuracy
ruMMLU 0.686 Accuracy
SimpleAr 0.996 Exact match
ruHumanEval 0.006 / 0.006 / 0.006 Pass@k
ruHHH 0.753
ruHateSpeech 0.755
ruDetox 0.293
ruEthics
Correct God Ethical
Virtue 0.446 0.401 0.525
Law 0.447 0.417 0.499
Moral 0.492 0.433 0.564
Justice 0.415 0.364 0.463
Utilitarianism 0.397 0.355 0.426

Information about the submission

Mera version
v1.2.0
Torch Version
2.8.0
The version of the codebase
2a537dd
CUDA version
12.8
Precision of the model weights
bfloat16
Seed
1234
Batch
4
Transformers version
4.57.1
The number of GPUs and their type
4 x NVIDIA A100 80GB PCIe
Architecture
vllm

Team:

Avito

Name of the ML model:

A-vibe

Model size

8.0B

Model type:

Opened

SFT

Architecture description:

https://habr.com/ru/companies/avito/articles/956664/

License:

https://www.apache.org/licenses/LICENSE-2.0

Inference parameters

Generation Parameters:
simplear - do_sample=false;until=[" \n"]; \nchegeka - do_sample=false;until=[" \n"]; \nrudetox - do_sample=false;until=[" \n"]; \nrumultiar - do_sample=false;until=[" \n"]; \nuse - do_sample=false;until=[" \n","."]; \nmultiq - do_sample=false;until=[" \n"]; \nrumodar - do_sample=false;until=[" \n"]; \nruhumaneval - do_sample=true;temperature=0.6;until=[" \nclass"," \ndef"," \n#"," \nif"," \nprint"]; \nrucodeeval - do_sample=true;temperature=0.6;until=[" \nclass"," \ndef"," \n#"," \nif"," \nprint"];do_sample=true;temperature=0.6;until=[" \nclass"," \ndef"," \n#"," \nif"," \nprint"];

System prompt:
Реши задачу по инструкции ниже. Не давай никаких объяснений и пояснений к своему ответу. Не пиши ничего лишнего. Пиши только то, что указано в инструкции. Если по инструкции нужно решить пример, то напиши только числовой ответ без хода решения и пояснений. Если по инструкции нужно вывести букву, цифру или слово, выведи только его. Если по инструкции нужно выбрать один из вариантов ответа и вывести букву или цифру, которая ему соответствует, то выведи только эту букву или цифру, не давай никаких пояснений, не добавляй знаки препинания, только 1 символ в ответе. Если по инструкции нужно дописать код функции на языке Python, начни ответ с перевода строки, затем поставь четыре пробела и потом пиши код будто ты продолжаешь функцию из инструкции, НИ В КОЕМ СЛУЧАЕ НЕ пиши сигнатуру функции в ответе, используй сигнатуру функции из задания, обязательно соблюдай отступы, не давай пояснений, не пиши комментарии, используй только аргументы из сигнатуры функции в инструкции, не пробуй считывать данные через функцию input. Не извиняйся, не строй диалог. Выдавай только ответ и ничего больше.

Description of the template:
{%- if tools %} \n {{- '<|im_start|>system \n' }} \n {%- if messages[0]['role'] == 'system' %} \n {{- messages[0]['content'] }} \n {%- else %} \n {{- '' }} \n {%- endif %} \n {{- " \n \n# Tools \n \nYou may call one or more functions to assist with the user query. \n \nYou are provided with function signatures within <tools></tools> XML tags: \n<tools>" }} \n {%- for tool in tools %} \n {{- " \n" }} \n {{- tool | tojson }} \n {%- endfor %} \n {{- " \n</tools> \n \nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags: \n<tool_call> \n{"name": <function-name>, "arguments": <args-json-object>} \n</tool_call><|im_end|> \n" }} \n{%- else %} \n {%- if messages[0]['role'] == 'system' %} \n {{- '<|im_start|>system \n' + messages[0]['content'] + '<|im_end|> \n' }} \n {%- endif %} \n{%- endif %} \n{%- for message in messages %} \n {%- if (message.role == "user") or (message.role == "system" and not loop.first) %} \n {{- '<|im_start|>' + message.role + ' \n' + message.content + '<|im_end|>' + ' \n' }} \n {%- elif message.role == "assistant" and not message.tool_calls %} \n {%- set content = message.content %} \n {%- if not loop.last %} \n {%- set content = message.content.split('</think>')[-1].lstrip(' \n') %} \n {%- endif %} \n {{- '<|im_start|>' + message.role + ' \n' + content + '<|im_end|>' + ' \n' }} \n {%- elif message.role == "assistant" %} \n {%- set content = message.content %} \n {%- if not loop.last %} \n {%- set content = message.content.split('</think>')[-1].lstrip(' \n') %} \n {%- endif %} \n {{- '<|im_start|>' + message.role }} \n {%- if message.content %} \n {{- ' \n' + content }} \n {%- endif %} \n {%- for tool_call in message.tool_calls %} \n {%- if tool_call.function is defined %} \n {%- set tool_call = tool_call.function %} \n {%- endif %} \n {{- ' \n<tool_call> \n{"name": "' }} \n {{- tool_call.name }} \n {{- '", "arguments": ' }} \n {%- if tool_call.arguments is string %} \n {{- tool_call.arguments }} \n {%- else %} \n {{- tool_call.arguments | tojson }} \n {%- endif %} \n {{- '} \n</tool_call>' }} \n {%- endfor %} \n {{- '<|im_end|> \n' }} \n {%- elif message.role == "tool" %} \n {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %} \n {{- '<|im_start|>user' }} \n {%- endif %} \n {{- ' \n<tool_response> \n' }} \n {{- message.content }} \n {{- ' \n</tool_response>' }} \n {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %} \n {{- '<|im_end|> \n' }} \n {%- endif %} \n {%- endif %} \n{%- endfor %} \n{%- if add_generation_prompt %} \n {{- '<|im_start|>assistant \n' }} \n{%- endif %}

Ratings by subcategory

Metric: Grade Norm
Model, team 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 8_0 8_1 8_2 8_3 8_4
A-vibe
Avito
0.733 0.467 0.767 0.067 0.1 0.5 0.133 - 0.1 0.1 0.033 0.067 0.2 0.033 0.133 0.567 0.167 0.167 0.433 0.067 0.3 0.7 0.533 0.2 0.2 0.583 0.433 0.467 0.667 0.667 0.7
Model, team Honest Helpful Harmless
A-vibe
Avito
0.754 0.678 0.828
Model, team Anatomy Virology Astronomy Marketing Nutrition Sociology Management Philosophy Prehistory Human aging Econometrics Formal logic Global facts Jurisprudence Miscellaneous Moral disputes Business ethics Biology (college) Physics (college) Human Sexuality Moral scenarios World religions Abstract algebra Medicine (college) Machine learning Medical genetics Professional law PR Security studies Chemistry (школьная) Computer security International law Logical fallacies Politics Clinical knowledge Conceptual_physics Math (college) Biology (high school) Physics (high school) Chemistry (high school) Geography (high school) Professional medicine Electrical engineering Elementary mathematics Psychology (high school) Statistics (high school) History (high school) Math (high school) Professional accounting Professional psychology Computer science (college) World history (high school) Macroeconomics Microeconomics Computer science (high school) European history Government and politics
A-vibe
Avito
0.607 0.506 0.822 0.842 0.725 0.811 0.806 0.72 0.691 0.659 0.57 0.611 0.45 0.741 0.746 0.676 0.74 0.854 0.578 0.756 0.563 0.795 0.55 0.728 0.571 0.79 0.469 0.639 0.747 0.48 0.75 0.769 0.693 0.798 0.721 0.761 0.56 0.881 0.642 0.734 0.793 0.676 0.69 0.817 0.86 0.699 0.77 0.585 0.525 0.656 0.62 0.814 0.767 0.828 0.83 0.812 0.782
Model, team SIM FL STA
A-vibe
Avito
0.784 0.64 0.622
Model, team Anatomy Virology Astronomy Marketing Nutrition Sociology Managment Philosophy Pre-History Gerontology Econometrics Formal logic Global facts Jurisprudence Miscellaneous Moral disputes Business ethics Bilology (college) Physics (college) Human sexuality Moral scenarios World religions Abstract algebra Medicine (college) Machine Learning Genetics Professional law PR Security Chemistry (college) Computer security International law Logical fallacies Politics Clinical knowledge Conceptual physics Math (college) Biology (high school) Physics (high school) Chemistry (high school) Geography (high school) Professional medicine Electrical Engineering Elementary mathematics Psychology (high school) Statistics (high school) History (high school) Math (high school) Professional Accounting Professional psychology Computer science (college) World history (high school) Macroeconomics Microeconomics Computer science (high school) Europe History Government and politics
A-vibe
Avito
0.667 0.881 0.717 0.639 0.697 0.69 0.724 0.737 0.615 0.708 0.769 0.783 0.508 0.798 0.69 0.642 0.757 0.756 0.702 0.825 0.737 0.78 0.867 0.84 0.844 0.742 0.744 0.702 0.912 0.889 0.844 0.821 0.75 0.842 0.697 0.768 0.867 0.822 0.737 0.754 0.82 0.825 0.844 0.956 0.897 0.844 0.897 0.909 0.738 0.947 0.822 0.725 0.823 0.701 0.628 0.585 0.789
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
A-vibe
Avito
0.446 0.447 0.492 0.415 0.397
Model, team Virtue Law Moral Justice Utilitarianism
A-vibe
Avito
0.401 0.417 0.433 0.364 0.355
Model, team Virtue Law Moral Justice Utilitarianism
A-vibe
Avito
0.525 0.499 0.564 0.463 0.426
Model, team Women Men LGBT Nationalities Migrants Other
A-vibe
Avito
0.741 0.657 0.706 0.703 0.714 0.885