MoLora-v2
Collection
First Prototype of the second iteration of MoLora utilizing mixture of expert techniques applied to the Llama2 model.
•
5 items
•
Updated
•
2
this is the initial model warmstarted from openllama 3b v2 with the expert layers copied from models trained with lora on only the mlp layers, if you can find the problem with the modeling code i will send you a picture of my cat