GigaCode 1.4

MERA Created at 15.07.2025 05:06

0.293

The overall result

Ratings for leaderboard tasks

The table will scroll to the left

Task name	Result	Metric
YABLoCo	0.135 / 0.038	EM pass@k
stRuCom	0.276	chrF
RealCode	0.322 / 0.988	pass@k execution_success
UnitTests	0.23	CodeBLEU
ruCodeEval	0.357 / 0.364 / 0.366	pass@k
JavaTestGen	0.313 / 0.639	pass@k compile@1
ruHumanEval	0.305 / 0.305 / 0.305	pass@k
RealCodeJava	0.352 / 0.997	pass@k execution_success
CodeLinterEval	0.027 / 0.027 / 0.027	pass@k
ruCodeReviewer	0.029 / 0.182 / 0.064 / 0.12 / 0.145	chrF BLEU judge@1 judge@5 judge@10
CodeCorrectness	0.676	EM

Information about the submission

Mera version

v.1.0.0

Torch Version

2.7.1

The version of the codebase

601ed3a

CUDA version

12.6

Precision of the model weights

auto

Seed

1234

Batch

Transformers version

4.53.0

The number of GPUs and their type

Architecture

local-chat-completions

Team:

MERA

Name of the ML model:

GigaCode 1.4

Link to the ML model:

https://gitverse.ru/features/gigacode/

Model size

32.0B

Model type:

Closed

API

SFT

Additional links:

Architecture description:

Description of the training:

Pretrain data:

License:

Proprietary model by Sber

GigaCode 1.4

Ratings for leaderboard tasks

Information about the submission

Team:

Name of the ML model:

Link to the ML model:

Model size

Model type:

Additional links:

Architecture description:

Description of the training:

Pretrain data:

License:

Confirm the deletion of the sub