SwissArmyTransformer
SwissArmyTransformer copied to clipboard
what is the order of modules inserted by `add_mix` in Transformer?
Hello, I would like to ask two questions as follows:
- what determines the location of the module inserted by
add_mix
? If not specified, is it inserted at the end? - why not save the zero optimizer ? Just to save disk space?
If any hooks of the mixins are conflicted, an error is raised, except the hook having a @non_conflict decorator. The mixins objects are saved at the model(.module if deepspeed).mixins, which is a ModuleDict.