This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

Updates paths inside resnets to the new naming scheme (local renaming)

Updates paths inside resnets to the new naming scheme (local renaming)

Updates paths inside resnets to the new naming scheme (local renaming)

Get an attribute off the `socket` module, or use a default.

Removes segments. Positive values shave the first segments, negative shave the last segments.

Removes segments. Positive values shave the first segments, negative shave the last segments.

Updates paths inside attentions to the new naming scheme (local renaming)

Updates paths inside resnets to the new naming scheme (local renaming)

Updates paths inside attentions to the new naming scheme (local renaming)

Updates paths inside attentions to the new naming scheme (local renaming)

Creates a config for the diffusers based on the config of the LDM model.

Invoke command-line Pip with the specified args. Returns: A bytestring containing the output of Pip.

Updates paths inside attentions to the new naming scheme (local renaming)

Decorator to register a function as a check. Checks are run in the order in which they are registered. Args: fn: A function that takes

Takes a state dict and a config, and returns a converted checkpoint.

Takes a state dict and a config, and returns a converted checkpoint.

Creates a config for the diffusers based on the config of the LDM model.

Creates a config for the diffusers based on the config of the LDM model.

Creates a config for the diffusers based on the config of the LDM model.

Return the path to a binary, or `None` if it's not on the path. Returns: A bytestring.

Print the Git hash of this version of `diagnose_tensorboard.py`. Given this hash, use `git cat-file blob HASH` to recover the relevant versio