ruadapt_qwen2.5_3B_ext_u48_instruct_v4

RCC MSU Created at 18.10.2024 12:57
0.412
The overall result
311
Place in the rating
Weak tasks:
384
RWSD
321
PARus
259
RCB
344
ruEthics
256
MultiQ
270
ruWorldTree
264
ruOpenBookQA
392
CheGeKa
278
ruMMLU
223
ruHateSpeech
223
ruDetox
278
ruHHH
182
ruTiE
323
ruHumanEval
336
USE
222
MathLogicQA
231
ruMultiAr
245
SimpleAr
345
LCS
326
BPS
420
ruModAr
274
MaMuRAMu
+18
Hide

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.094 Accuracy
RCB 0.53 / 0.34 Accuracy F1 macro
USE 0.109 Grade norm
RWSD 0.504 Accuracy
PARus 0.732 Accuracy
ruTiE 0.728 Accuracy
MultiQ 0.368 / 0.23 F1 Exact match
CheGeKa 0.035 / 0.019 F1 Exact match
ruModAr 0.283 Exact match
MaMuRAMu 0.637 Accuracy
ruMultiAr 0.277 Exact match
ruCodeEval 0 / 0 / 0 Pass@k
MathLogicQA 0.432 Accuracy
ruWorldTree 0.859 / 0.858 Accuracy F1 macro
ruOpenBookQA 0.77 / 0.767 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 0.892 Accuracy
ruMMLU 0.533 Accuracy
SimpleAr 0.962 Exact match
ruHumanEval 0.005 / 0.012 / 0.012 Pass@k
ruHHH 0.646
ruHateSpeech 0.747
ruDetox 0.197
ruEthics
Correct God Ethical
Virtue 0.238 0.203 0.175
Law 0.214 0.179 0.169
Moral 0.241 0.216 0.185
Justice 0.203 0.168 0.149
Utilitarianism 0.184 0.167 0.177

Information about the submission

Mera version
v.1.2.0
Torch Version
2.4.0
The version of the codebase
430295f
CUDA version
12.1
Precision of the model weights
bfloat16
Seed
1234
Batch
1
Transformers version
4.45.2
The number of GPUs and their type
NVIDIA A100-SXM4-80GB
Architecture
vllm

Team:

RCC MSU

Name of the ML model:

ruadapt_qwen2.5_3B_ext_u48_instruct_v4

Model size

3.0B

Model type:

Opened

Pretrain

SFT

Architecture description:

Tikhomirov M., Chernyshev D. Facilitating large language model Russian adaptation with Learned Embedding Propagation // 2024 (will be soon) Tikhomirov M., Chernyshev D. Impact of Tokenization on LLaMa Russian Adaptation //2023 Ivannikov Ispras Open Conference (ISPRAS). – IEEE, 2023. – С. 163-168.

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, chegeka, rudetox, rumultiar, use, multiq, rumodar, ruhumaneval, rucodeeval - 32768 \nrutie - 8192

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 {%- 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 %}

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
ruadapt_qwen2.5_3B_ext_u48_instruct_v4
RCC MSU
0.533 0.033 0.633 0.067 0 0.1 0 - 0 0 0.033 0.033 0 0 0.1 0.083 0 0.033 0 0 0 0.3 0 0.033 0 0.242 0 0.1 0.2 0.167 0.2
Model, team Honest Helpful Harmless
ruadapt_qwen2.5_3B_ext_u48_instruct_v4
RCC MSU
0.639 0.644 0.655
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
ruadapt_qwen2.5_3B_ext_u48_instruct_v4
RCC MSU
0.4 0.392 0.625 0.735 0.585 0.711 0.68 0.601 0.593 0.61 0.43 0.373 0.29 0.676 0.622 0.555 0.67 0.514 0.322 0.588 0.359 0.708 0.33 0.566 0.33 0.55 0.366 0.509 0.665 0.43 0.7 0.727 0.509 0.758 0.551 0.526 0.4 0.671 0.424 0.463 0.712 0.482 0.572 0.554 0.697 0.528 0.618 0.419 0.379 0.482 0.46 0.764 0.564 0.534 0.65 0.752 0.606
Model, team SIM FL STA
ruadapt_qwen2.5_3B_ext_u48_instruct_v4
RCC MSU
0.707 0.641 0.495
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
ruadapt_qwen2.5_3B_ext_u48_instruct_v4
RCC MSU
0.444 0.663 0.583 0.63 0.671 0.707 0.603 0.649 0.558 0.585 0.679 0.6 0.475 0.659 0.556 0.593 0.664 0.689 0.474 0.754 0.404 0.78 0.667 0.633 0.622 0.606 0.654 0.561 0.86 0.733 0.733 0.756 0.563 0.754 0.545 0.679 0.622 0.756 0.596 0.569 0.657 0.683 0.778 0.756 0.897 0.889 0.69 0.795 0.708 0.789 0.778 0.609 0.709 0.481 0.395 0.456 0.722
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
ruadapt_qwen2.5_3B_ext_u48_instruct_v4
RCC MSU
0.238 0.214 0.241 0.203 0.184
Model, team Virtue Law Moral Justice Utilitarianism
ruadapt_qwen2.5_3B_ext_u48_instruct_v4
RCC MSU
0.203 0.179 0.216 0.168 0.167
Model, team Virtue Law Moral Justice Utilitarianism
ruadapt_qwen2.5_3B_ext_u48_instruct_v4
RCC MSU
0.175 0.169 0.185 0.149 0.177
Model, team Women Men LGBT Nationalities Migrants Other
ruadapt_qwen2.5_3B_ext_u48_instruct_v4
RCC MSU
0.778 0.686 0.588 0.757 0.571 0.787