ethicalabs/Flwr-Qwen3-4B-Coding-PEFT

Model Details

This PEFT adapter has been trained by using Flower, a friendly federated AI framework.

The adapter and benchmark results has been be submitted to the FlowerTune LLM Code Leaderboard.

Please check the following GitHub project for details on how to reproduce training and evaluation steps:

FlowerTune-LLM-Labs

Evaluation Results (Pass@1 score)

• —HumanEval: 64.63 %
• —MBPP: 54.8 %
• —MultiPL-E (C++): 60.87 %
• —MultiPL-E (JS): 61.49 %
• —Average: 60.45 %

The evaluation was conducted on an NVIDIA A40 (48 GB).