GigaChat 2 Pro

GIGACHAT Created at 12.03.2025 11:33
0.649
The overall result
43
Place in the rating
Weak tasks:
47
RWSD
34
PARus
127
RCB
241
ruEthics
118
MultiQ
26
ruWorldTree
53
ruOpenBookQA
129
CheGeKa
79
ruMMLU
344
ruHateSpeech
333
ruDetox
95
ruHHH
28
ruTiE
20
ruHumanEval
26
USE
47
MathLogicQA
78
ruMultiAr
65
SimpleAr
154
LCS
250
BPS
31
ruModAr
65
MaMuRAMu
23
ruCodeEval
+19
Hide

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.138 Accuracy
RCB 0.562 / 0.392 Accuracy F1 macro
USE 0.534 Grade norm
RWSD 0.665 Accuracy
PARus 0.934 Accuracy
ruTiE 0.883 Accuracy
MultiQ 0.492 / 0.37 F1 Exact match
CheGeKa 0.296 / 0.248 F1 Exact match
ruModAr 0.943 Exact match
MaMuRAMu 0.831 Accuracy
ruMultiAr 0.396 Exact match
ruCodeEval 0.599 / 0.633 / 0.64 Pass@k
MathLogicQA 0.775 Accuracy
ruWorldTree 0.99 / 0.99 Accuracy F1 macro
ruOpenBookQA 0.93 / 0.746 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 0.938 Accuracy
ruMMLU 0.745 Accuracy
SimpleAr 0.994 Exact match
ruHumanEval 0.627 / 0.672 / 0.683 Pass@k
ruHHH 0.837
ruHateSpeech 0.634
ruDetox 0.142
ruEthics
Correct God Ethical
Virtue 0.273 0.283 0.355
Law 0.274 0.276 0.34
Moral 0.297 0.297 0.376
Justice 0.236 0.249 0.311
Utilitarianism 0.229 0.251 0.29

Information about the submission

Mera version
v.1.2.0
Torch Version
2.5.1
The version of the codebase
30667dc
CUDA version
12.4
Precision of the model weights
-
Seed
1234
Batch
1
Transformers version
4.49.0
The number of GPUs and their type
-
Architecture
gigachat_llms

Team:

GIGACHAT

Name of the ML model:

GigaChat 2 Pro

Model type:

Closed

API

SFT

Architecture description:

GigaChat 2 Pro is a new Large Language Model (LLM) by Sber. The version is available for users via API in preview mode https://developers.sber.ru/portal/products/gigachat-api

Description of the training:

-

Pretrain data:

-

License:

Proprietary model by Sber

Inference parameters

Generation Parameters:
rucodeeval - do_sample=true;temperature=0.6;until=["\nclass","\ndef","\n#","\nif","\nprint"]; \nruhumaneval - 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, 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:
API Default

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
GigaChat 2 Pro
GIGACHAT
0.833 0.6 0.833 0.3 0.433 0.767 0.4 - 0.367 0.167 0.167 0.2 0.633 0.267 0.1 0.683 0.267 0.3 0.567 0.167 0.6 0.733 0.6 0.433 0.367 0.767 0.7 0.6 0.8 0.7 0.833
Model, team Honest Helpful Harmless
GigaChat 2 Pro
GIGACHAT
0.803 0.814 0.897
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
GigaChat 2 Pro
GIGACHAT
0.711 0.584 0.901 0.846 0.837 0.861 0.816 0.749 0.87 0.74 0.64 0.603 0.55 0.759 0.842 0.746 0.8 0.896 0.644 0.794 0.647 0.848 0.58 0.728 0.679 0.85 0.491 0.667 0.759 0.58 0.8 0.843 0.736 0.919 0.781 0.876 0.66 0.91 0.715 0.773 0.889 0.783 0.738 0.854 0.886 0.722 0.819 0.607 0.539 0.738 0.77 0.793 0.854 0.92 0.91 0.794 0.896
Model, team SIM FL STA
GigaChat 2 Pro
GIGACHAT
0.236 0.725 0.767
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
GigaChat 2 Pro
GIGACHAT
0.689 0.901 0.8 0.713 0.895 0.845 0.672 0.737 0.865 0.785 0.808 0.8 0.492 0.845 0.789 0.778 0.785 0.844 0.825 0.86 0.86 0.949 0.933 0.905 0.822 0.909 0.885 0.754 0.912 0.844 0.844 0.91 0.786 0.86 0.758 0.857 0.956 0.844 0.789 0.785 0.878 0.905 0.867 0.978 0.931 0.933 0.897 0.932 0.892 0.93 0.844 0.855 0.823 0.766 0.721 0.83 0.844
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
GigaChat 2 Pro
GIGACHAT
0.273 0.274 0.297 0.236 0.229
Model, team Virtue Law Moral Justice Utilitarianism
GigaChat 2 Pro
GIGACHAT
0.283 0.276 0.297 0.249 0.251
Model, team Virtue Law Moral Justice Utilitarianism
GigaChat 2 Pro
GIGACHAT
0.355 0.34 0.376 0.311 0.29
Model, team Women Men LGBT Nationalities Migrants Other
GigaChat 2 Pro
GIGACHAT
0.843 0.6 0 0.135 0.286 0.803