Qwen3-4B

MERA Created at 04.03.2026 21:29
0.671
The overall result
38
Place in the rating
Weak tasks:
79
RWSD
187
PARus
203
RCB
40
ruEthics
244
MultiQ
106
ruWorldTree
170
ruOpenBookQA
285
CheGeKa
107
ruMMLU
129
ruHateSpeech
140
ruDetox
133
ruHHH
105
ruTiE
86
USE
23
ruMultiAr
137
SimpleAr
30
LCS
263
BPS
155
MaMuRAMu
20
ruCodeEval
+16
Hide

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.366 Accuracy
RCB 0.543 / 0.372 Accuracy F1 macro
USE 0.317 Grade norm
RWSD 0.615 Accuracy
PARus 0.86 Accuracy
ruTiE 0.806 Accuracy
MultiQ 0.379 / 0.236 F1 Exact match
CheGeKa 0.1 / 0.055 F1 Exact match
ruModAr 0.996 Exact match
MaMuRAMu 0.743 Accuracy
ruMultiAr 0.985 Exact match
ruCodeEval 0.648 / 0.742 / 0.762 Pass@k
MathLogicQA 0.994 Accuracy
ruWorldTree 0.964 / 0.964 Accuracy F1 macro
ruOpenBookQA 0.855 / 0.855 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 0.931 Accuracy
ruMMLU 0.704 Accuracy
SimpleAr 0.984 Exact match
ruHumanEval 0.762 / 0.842 / 0.86 Pass@k
ruHHH 0.798
ruHateSpeech 0.792
ruDetox 0.246
ruEthics
Correct God Ethical
Virtue 0.449 0.422 0.561
Law 0.451 0.414 0.537
Moral 0.47 0.455 0.598
Justice 0.423 0.371 0.489
Utilitarianism 0.364 0.368 0.489

Information about the submission

Mera version
v1.2.0
Torch Version
2.9.1
The version of the codebase
7030e16
CUDA version
12.8
Precision of the model weights
bfloat16
Seed
1234
Batch
1
Transformers version
4.56.1
The number of GPUs and their type
1 x NVIDIA A100-SXM4-80GB
Architecture
local-chat-completions

Team:

MERA

Name of the ML model:

Qwen3-4B

Link to the ML model:

https://huggingface.co/Qwen/Qwen3-4B

Model size

4.0B

Model type:

Opened

SFT

Additional links:

https://arxiv.org/abs/2505.09388

Architecture description:

Qwen3-4B is a decoder-only transformer language model from the Qwen3 family with approximately 4 billion parameters. The model is designed for general language understanding and generation tasks, including reasoning, coding, and multilingual applications, and supports long-context inputs.

Description of the training:

The model is trained using a multi-stage pipeline consisting of large-scale pretraining followed by post-training. Post-training includes instruction tuning and alignment techniques to improve instruction following, reasoning ability, and response quality.

Pretrain data:

The model was pretrained on a large multilingual corpus of approximately 36 trillion tokens covering 119 languages. It was further post-trained on instruction-following and reasoning-oriented datasets to improve instruction following and reasoning performance.

License:

Apache License 2.0

Inference parameters

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, пиши сразу код, соблюдая отступы так, будто ты продолжаешь функцию из инструкции, 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 %} {{- '<|im_start|>system \n' }} {%- if messages[0].role == 'system' %} {{- messages[0].content + ' \n \n' }} {%- endif %} {{- "# 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>" }} {%- for tool in tools %} {{- " \n" }} {{- tool | tojson }} {%- endfor %} {{- " \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" }} {%- else %} {%- if messages[0].role == 'system' %} {{- '<|im_start|>system \n' + messages[0].content + '<|im_end|> \n' }} {%- endif %} {%- endif %} {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %} {%- for message in messages[::-1] %} {%- set index = (messages|length - 1) - loop.index0 %} {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %} {%- set ns.multi_step_tool = false %} {%- set ns.last_query_index = index %} {%- endif %} {%- endfor %} {%- for message in messages %} {%- if message.content is string %} {%- set content = message.content %} {%- else %} {%- set content = '' %} {%- endif %} {%- if (message.role == "user") or (message.role == "system" and not loop.first) %} {{- '<|im_start|>' + message.role + ' \n' + content + '<|im_end|>' + ' \n' }} {%- elif message.role == "assistant" %} {%- set reasoning_content = '' %} {%- if message.reasoning_content is string %} {%- set reasoning_content = message.reasoning_content %} {%- else %} {%- if '</think>' in content %} {%- set reasoning_content = content.split('</think>')[0].rstrip(' \n').split('<think>')[-1].lstrip(' \n') %} {%- set content = content.split('</think>')[-1].lstrip(' \n') %} {%- endif %} {%- endif %} {%- if loop.index0 > ns.last_query_index %} {%- if loop.last or (not loop.last and reasoning_content) %} {{- '<|im_start|>' + message.role + ' \n<think> \n' + reasoning_content.strip(' \n') + ' \n</think> \n \n' + content.lstrip(' \n') }} {%- else %} {{- '<|im_start|>' + message.role + ' \n' + content }} {%- endif %} {%- else %} {{- '<|im_start|>' + message.role + ' \n' + content }} {%- endif %} {%- if message.tool_calls %} {%- for tool_call in message.tool_calls %} {%- if (loop.first and content) or (not loop.first) %} {{- ' \n' }} {%- endif %} {%- if tool_call.function %} {%- set tool_call = tool_call.function %} {%- endif %} {{- '<tool_call> \n{"name": "' }} {{- tool_call.name }} {{- '", "arguments": ' }} {%- if tool_call.arguments is string %} {{- tool_call.arguments }} {%- else %} {{- tool_call.arguments | tojson }} {%- endif %} {{- '} \n</tool_call>' }} {%- endfor %} {%- endif %} {{- '<|im_end|> \n' }} {%- elif message.role == "tool" %} {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %} {{- '<|im_start|>user' }} {%- endif %} {{- ' \n<tool_response> \n' }} {{- content }} {{- ' \n</tool_response>' }} {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %} {{- '<|im_end|> \n' }} {%- endif %} {%- endif %} {%- endfor %} {%- if add_generation_prompt %} {{- '<|im_start|>assistant \n' }} {%- if enable_thinking is defined and enable_thinking is false %} {{- '<think> \n \n</think> \n \n' }} {%- endif %} {%- 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
Qwen3-4B
MERA
0.5 0.333 0.833 0.267 0.133 0.333 0.033 - 0.1 0.067 0.067 0.133 0.233 0.233 0.167 0.367 0.067 0.067 0.067 0.033 0.3 0.633 0.267 0.333 0.233 0.575 0.333 0.533 0.333 0.567 0.533
Model, team Honest Helpful Harmless
Qwen3-4B
MERA
0.754 0.763 0.879
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
Qwen3-4B
MERA
0.674 0.494 0.855 0.812 0.732 0.776 0.806 0.659 0.685 0.646 0.658 0.762 0.49 0.704 0.723 0.633 0.69 0.813 0.889 0.779 0.442 0.719 0.82 0.717 0.759 0.72 0.448 0.583 0.665 0.67 0.72 0.736 0.699 0.758 0.751 0.846 0.83 0.874 0.841 0.872 0.818 0.761 0.724 0.939 0.837 0.833 0.76 0.974 0.628 0.666 0.8 0.781 0.815 0.878 0.9 0.764 0.808
Model, team SIM FL STA
Qwen3-4B
MERA
0.715 0.496 0.728
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
Qwen3-4B
MERA
0.556 0.842 0.667 0.611 0.803 0.672 0.621 0.614 0.635 0.677 0.782 0.758 0.483 0.698 0.608 0.691 0.785 0.778 0.772 0.807 0.509 0.78 0.933 0.775 0.911 0.773 0.718 0.667 0.877 0.822 0.822 0.808 0.777 0.825 0.606 0.821 0.911 0.844 0.86 0.846 0.759 0.794 0.844 1 0.879 0.911 0.931 0.932 0.708 0.807 0.889 0.638 0.747 0.714 0.674 0.602 0.789
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
Qwen3-4B
MERA
0.449 0.451 0.47 0.423 0.364
Model, team Virtue Law Moral Justice Utilitarianism
Qwen3-4B
MERA
0.422 0.414 0.455 0.371 0.368
Model, team Virtue Law Moral Justice Utilitarianism
Qwen3-4B
MERA
0.561 0.537 0.598 0.489 0.489
Model, team Women Men LGBT Nationalities Migrants Other
Qwen3-4B
MERA
0.806 0.771 0.824 0.73 0.571 0.836