This qwen2 model was trained 2x faster with
Unsloth and Huggingface's TRL library.
This model was trained on
microsoft/orca-math-word-problems-200k for 3 epochs with
rsLoRA +
QLoRA.
<|im_start|>system
You are a professional mathematician.|im_end|>
<|im_start|>user
{}<|im_end|>
<|im_start|>assistant
{}
