MEGAN Coder 7B
MEGAN Coder 7B is a QLoRA adapter trained from:
Base model: Qwen/Qwen2.5-Coder-7B-Instruct
Training dataset: flwrlabs/code-alpaca-20k
This repository contains parameter-efficient adapter weights rather than a copy of the full base model.
Stage
Stage 1 — coding instruction specialization.
Training
The adapter was trained using 4-bit QLoRA with PEFT.
Provenance
The exact base model and dataset revisions are recorded in
megan_run_metadata.json.
Usage
Load the base model separately and attach the adapter using PEFT.
- Downloads last month
- -