Kodify Nano

MERA Created at 16.07.2025 08:57
0.181
The overall result

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
YABLoCo 0.082 / 0
EM pass@k
stRuCom 0.092
chrF
RealCode 0.041 / 0.988
pass@k execution_success
UnitTests 0.112
CodeBLEU
ruCodeEval 0.191 / 0.313 / 0.354
pass@k
JavaTestGen 0.053 / 0.3
pass@k compile@1
ruHumanEval 0.173 / 0.299 / 0.341
pass@k
RealCodeJava 0.104 / 0.97
pass@k execution_success
CodeLinterEval 0.235 / 0.432 / 0.473
pass@k
ruCodeReviewer 0.005 / 0.043 / 0.003 / 0.003 / 0.003
chrF BLEU judge@1 judge@5 judge@10
CodeCorrectness 0.475
EM

Information about the submission

Mera version
v.1.0.0
Torch Version
2.7.0
The version of the codebase
e77f33f
CUDA version
12.6
Precision of the model weights
auto
Seed
1234
Batch
1
Transformers version
4.53.0
The number of GPUs and their type
1 x NVIDIA A100-SXM4-40GB
Architecture
vllm

Team:

MERA

Name of the ML model:

Kodify Nano

Model size

1.5B

Model type:

Opened

SFT

Additional links:

-

Architecture description:

Kodify-Nano – это легковесная LLM, разработанная для выполнения задач разработки кода с минимальными ресурсами. Она оптимизирована для быстрого и эффективного взаимодействия с пользователями, обеспечивая высокую производительность даже в условиях ограниченных ресурсов. Kodify-Nano is a lightweight LLM designed for code development tasks with minimal resource usage. It is optimized for fast and efficient interaction, delivering high performance even in resource-constrained environments.

Description of the training:

-

Pretrain data:

-

License:

https://huggingface.co/MTSAIR/Kodify-Nano/blob/main/Apache%20License%20MTS%20AI.docx