fix error of parameter initialization for LoRA embedding

[hero007feng]

I found that the parameter initialization at reset_parameters() of the Embedding class differs from the LoRa paper and other implementations at layers.py. I initialized lora_A with nn.init.normal_() while Lora_B with nn.init.zeros_(). Thanks.

[edwardjhu]

Thanks! This was also pointed out by someone else in an issue. I am inclined to keep it as it is for reproducibility since it's functionally equivalent to what is written in the paper. However, please let me know if this discrepancy is causing any issues!