Qwen2.5-3B-Instruct

MERA Created at 04.11.2024 17:46
0.415
The overall result
255
Place in the rating

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.088 Accuracy
RCB 0.548 / 0.357 Accuracy F1 macro
USE 0.147 Grade norm
RWSD 0.462 Accuracy
PARus 0.766 Accuracy
ruTiE 0.721 Accuracy
MultiQ 0.391 / 0.263 F1 Exact match
CheGeKa 0.029 / 0.017 F1 Exact match
ruModAr 0.318 Exact match
MaMuRAMu 0.632 Accuracy
ruMultiAr 0.279 Exact match
ruCodeEval 0 / 0 / 0 Pass@k
MathLogicQA 0.428 Accuracy
ruWorldTree 0.861 / 0.861 Accuracy F1 macro
ruOpenBookQA 0.72 / 0.718 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 0.905 Accuracy
ruMMLU 0.516 Accuracy
SimpleAr 0.965 Exact match
ruHumanEval 0.001 / 0.005 / 0.006 Pass@k
ruHHH 0.657
ruHateSpeech 0.747
ruDetox 0.173
ruEthics
Correct God Ethical
Virtue 0.153 0.203 0.191
Law 0.149 0.182 0.173
Moral 0.167 0.204 0.2
Justice 0.114 0.172 0.153
Utilitarianism 0.119 0.187 0.193

Information about the submission:

Mera version
v.1.2.0
Torch Version
2.4.0
The version of the codebase
9b26db97
CUDA version
12.1
Precision of the model weights
bfloat16
Seed
1234
Butch
1
Transformers version
4.44.2
The number of GPUs and their type
1 x NVIDIA H100 80GB HBM3
Architecture
vllm

Team:

MERA

Name of the ML model:

Qwen2.5-3B-Instruct

Model size

3.0B

Model type:

Opened

SFT

Additional links:

https://qwenlm.github.io/blog/qwen2.5-llm/

Architecture description:

Type: Causal Language Models Training Stage: Pretraining & Post-training Architecture: transformers with RoPE, SwiGLU, RMSNorm, Attention QKV bias and tied word embeddings Number of Parameters: 3.09B Number of Paramaters (Non-Embedding): 2.77B Number of Layers: 36 Number of Attention Heads (GQA): 16 for Q and 2 for KV Context Length: Full 32,768 tokens and generation 8192 tokens

Description of the training:

-

Pretrain data:

The size of the pre-training dataset is expanded from 7 trillion tokens to a maximum of 18 trillion tokens.

License:

qwen-research

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;until=["\nclass","\ndef","\n#","\nif","\nprint"];temperature=0.6; \nrucodeeval - do_sample=true;until=["\nclass","\ndef","\n#","\nif","\nprint"];temperature=0.6;

The size of the context:
simplear, bps, lcs, chegeka, mathlogicqa, parus, rcb, rudetox, ruhatespeech, rummlu, ruworldtree, ruopenbookqa, rumultiar, use, rwsd, mamuramu, multiq, rumodar, ruethics, ruhhh, ruhumaneval, rucodeeval - 32768 \nrutie - 10000

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 %} \n {{- '<|im_start|>system\n' }} \n {%- if messages[0]['role'] == 'system' %} \n {{- messages[0]['content'] }} \n {%- else %} \n {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }} \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 {%- else %} \n {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|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) or (message.role == "assistant" and not message.tool_calls) %} \n {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }} \n {%- elif message.role == "assistant" %} \n {{- '<|im_start|>' + message.role }} \n {%- if message.content %} \n {{- '\n' + message.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 {{- tool_call.arguments | tojson }} \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 %}

Expand information

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
Qwen2.5-3B-Instruct
MERA
0.433 0.1 0.633 0.2 0 0.1 0 - 0.067 0.133 0.067 0 0 0 0.1 0.317 0 0 0.033 0 0 0.367 0.067 0 0 0.283 0.1 0.2 0.133 0.167 0.333
Model, team Honest Helpful Harmless
Qwen2.5-3B-Instruct
MERA
0.607 0.661 0.707
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
Qwen2.5-3B-Instruct
MERA
0.511 0.434 0.572 0.722 0.585 0.667 0.621 0.592 0.552 0.552 0.36 0.381 0.37 0.667 0.622 0.555 0.54 0.472 0.4 0.58 0.258 0.661 0.36 0.526 0.348 0.58 0.363 0.583 0.616 0.39 0.7 0.653 0.552 0.717 0.513 0.517 0.38 0.677 0.437 0.478 0.697 0.504 0.572 0.509 0.68 0.519 0.613 0.396 0.365 0.492 0.45 0.717 0.541 0.529 0.67 0.721 0.57
Model, team SIM FL STA
Qwen2.5-3B-Instruct
MERA
0.748 0.602 0.429
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
Qwen2.5-3B-Instruct
MERA
0.422 0.673 0.55 0.509 0.697 0.621 0.552 0.649 0.654 0.6 0.705 0.642 0.5 0.643 0.556 0.605 0.589 0.689 0.491 0.772 0.333 0.797 0.711 0.615 0.711 0.561 0.667 0.632 0.877 0.622 0.733 0.718 0.643 0.632 0.576 0.696 0.622 0.778 0.596 0.569 0.653 0.683 0.778 0.733 0.914 0.911 0.69 0.773 0.677 0.789 0.711 0.478 0.696 0.532 0.512 0.433 0.711
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
Qwen2.5-3B-Instruct
MERA
0.153 0.149 0.167 0.114 0.119
Model, team Virtue Law Moral Justice Utilitarianism
Qwen2.5-3B-Instruct
MERA
0.203 0.182 0.204 0.172 0.187
Model, team Virtue Law Moral Justice Utilitarianism
Qwen2.5-3B-Instruct
MERA
0.191 0.173 0.2 0.153 0.193
Model, team Women Men LGBT Nationalities Migrants Other
Qwen2.5-3B-Instruct
MERA
0.741 0.714 0.706 0.757 0.429 0.82