Text Industrial Code
Leaderboard Tasks How MERA works
Return to the leaderboard

GigaChat 2 Max

MERA Created at 15.07.2025 04:58
0.346
The overall result

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
YABLoCo 0.106 / 0.014
EM pass@k
stRuCom 0.294
chrF
RealCode 0.332 / 0.988
pass@k execution_success
UnitTests 0.175
CodeBLEU
ruCodeEval 0.537 / 0.588 / 0.591
pass@k
JavaTestGen 0.137 / 0.396
pass@k compile@1
ruHumanEval 0.53 / 0.57 / 0.579
pass@k
RealCodeJava 0.342 / 0.993
pass@k execution_success
CodeLinterEval 0.425 / 0.461 / 0.473
pass@k
ruCodeReviewer 0.016 / 0.136 / 0.062 / 0.065 / 0.071
chrF BLEU judge@1 judge@5 judge@10
CodeCorrectness 0.68
EM

Information about the submission:

Mera version
v.1.0.0
Torch Version
2.7.1
The version of the codebase
e77f33f
CUDA version
None
Precision of the model weights
auto
Seed
1234
Batch
1
Transformers version
4.53.0
The number of GPUs and their type
0
Architecture
gigachat-completion

Team:

MERA

Name of the ML model:

GigaChat 2 Max

Link to the ML model:

https://developers.sber.ru/docs/ru/gigachat/models

Model type:

Closed

API

SFT

Additional links:

-

Architecture description:

GigaChat 2 Max is a new Large Language Model (LLM) by Sber. The model will be available for B2C users https://giga.chat/ and B2B users via API soon https://developers.sber.ru/docs/ru/gigachat/api/tariffs.

Description of the training:

-

Pretrain data:

-

License:

Proprietary model by Sber

MERA

Useful links

© 2025 MERA. All rights reserved.
User agreement Privacy policy

We use cookies to store data. By continuing to use the site, you consent to work with these files