GigaChat-Pro

GIGACHAT Created at 04.07.2024 10:42

0.537

The overall result

The submission does not contain all the required tasks

Ratings for leaderboard tasks

The table will scroll to the left

Task name	Result	Metric
LCS	0.09	Accuracy
RCB	0.53 / 0.449	Accuracy F1 macro
USE	0.338	Grade norm
RWSD	0.585	Accuracy
PARus	0.884	Accuracy
ruTiE	0.791	Accuracy
MultiQ	0.369 / 0.247	F1 Exact match
CheGeKa	0.104 / 0	F1 Exact match
ruModAr	0.866	Exact match
ruMultiAr	0.273	Exact match
MathLogicQA	0.467	Accuracy
ruWorldTree	0.939 / 0.939	Accuracy F1 macro
ruOpenBookQA	0.873 / 0.872	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

0.318

Accuracy

ruMMLU

0.816

Accuracy

SimpleAr

0.971

Exact match

ruHumanEval

0.013 / 0.064 / 0.128

Pass@k

ruHHH

0.764

ruHateSpeech

0.751

ruDetox

0.238

ruEthics

	Correct	God	Ethical
Virtue	-0.493	-0.449	-0.394
Law	-0.493	-0.423	-0.392
Moral	-0.492	-0.464	-0.399
Justice	-0.447	-0.4	-0.345
Utilitarianism	-0.422	-0.374	-0.322

Information about the submission

Mera version

Torch Version

The version of the codebase

CUDA version

Precision of the model weights

Seed

Batch

Transformers version

The number of GPUs and their type

Architecture

Team:

GIGACHAT

Name of the ML model:

GigaChat-Pro

Link to the ML model:

https://developers.sber.ru/portal/products/gigachat-api

Additional links:

https://developers.sber.ru/docs/ru/gigachat/api/overview

Architecture description:

GigaChat Pro (version 1.0.26.8) is a Large Language Model (LLM) with 30B parameters that was fine-tuned on instruction corpus and has context length of 8192 tokens. The version is available for users via API since 13.07.

Description of the training:

Pretrain data:

Training Details:

License:

Proprietary model by Sber

Strategy, generation and parameters:

Code version v.1.1.0. All the parameters were not changed and are used as prepared by the organizers. Details: - 2 x NVIDIA A100 + accelerate - dtype float16 - Pytorch 2.3.1 + CUDA 12.1 - Transformers 4.42.3 - Context length 8192

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-Pro GIGACHAT	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-

Model, team	Honest	Helpful	Harmless
GigaChat-Pro GIGACHAT	0.689	0.78	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
GigaChat-Pro GIGACHAT	0.9	0.938	0.8	0.657	0.952	1	0.867	0.647	0.7	1	0.818	0.9	0.7	0.769	0.682	0.6	0.8	0.889	0.9	1	0.4	0.846	0.9	0.863	0.9	0.727	0.75	0.643	1	0.818	0.6	0.944	0.7	1	0.818	1	0.8	0.905	0.7	0.6	0.899	1	1	0.5	1	0.9	1	0.3	0.9	0.9	0.636	0.875	0.941	0.867	0.583	0.727	0.778

Model, team	SIM	FL	STA
GigaChat-Pro GIGACHAT	0.59	0.76	0.459

Coorect

Good

Ethical

Model, team	Virtue	Law	Moral	Justice	Utilitarianism
GigaChat-Pro GIGACHAT	-0.493	-0.493	-0.492	-0.447	-0.422

Model, team	Virtue	Law	Moral	Justice	Utilitarianism
GigaChat-Pro GIGACHAT	-0.449	-0.423	-0.464	-0.4	-0.374

Model, team	Virtue	Law	Moral	Justice	Utilitarianism
GigaChat-Pro GIGACHAT	-0.394	-0.392	-0.399	-0.345	-0.322

Model, team	Women	Men	LGBT	Nationalities	Migrants	Other
GigaChat-Pro GIGACHAT	0.759	0.8	0.647	0.649	0.429	0.836

GigaChat-Pro

Ratings for leaderboard tasks

Evaluation on open tasks:

Information about the submission

Team:

Name of the ML model:

Link to the ML model:

Additional links:

Architecture description:

Description of the training:

Pretrain data:

Training Details:

License:

Strategy, generation and parameters:

Ratings by subcategory

Confirm the deletion of the sub