dmx.compressor.fx.transformer.dmx_aware_transformer.DMXAwareTransformer

class dmx.compressor.fx.transformer.dmx_aware_transformer.DMXAwareTransformer(module: GraphModule, node_name_to_scope: dict, old_gms: List[GraphModule] = None, nodeInputs: dict = None)

Substitute as in dmx.model.aware(), replace torch.nn.modules and activations with dmx counterpart

Parameters:

• module (fx.GraphModule) – the module to transform

• node_name_to_scope (dict) – A dictionary storing the mapping between node names and scopes

module

the module to transform

Type:

fx.GraphModule

node_name_to_scope

A dictionary storing the mapping between node names and scopes

Type:

dict

old_gms List[

dmxmodules to reuse if module was already transformed

Type:

fx.GraphModule

__init__(module: GraphModule, node_name_to_scope: dict, old_gms: List[GraphModule] = None, nodeInputs: dict = None)

Note

Backwards-compatibility for this API is guaranteed.

Methods

__init__(module, node_name_to_scope[, ...])
add_dmx_aware_functional_mapping(target, ...)
add_submod(name, mod)	this function will try to reuse modules in old_gm if possible
boxed_run(args_list)	Run module via interpretation and return the result.
call_function(target, args, kwargs)	Execute a call_function node, replaces the function with its dmx counterpart and returns the result.
call_method(target, args, kwargs)	Execute a call_method node and return the result.
call_module(target, args, kwargs)	Execute a call_module node, replaces the module with its dmx counterpart and returns the result.
create_unique_name_in_scope(cand_name)
fetch_args_kwargs_from_env(n)	Fetch the concrete values of args and kwargs of node n from the current execution environment.
fetch_attr(target)	Fetch an attribute from the Module hierarchy of self.module.
get_attr(target, args, kwargs)	Execute a get_attr node.
map_nodes_to_values(args, n)	Recursively descend through args and look up the concrete value for each Node in the current execution environment.
merge_graph(subgraph, scope, args, kwargs)	This function is to merge subgraph to graph
output(target, args, kwargs)	Execute an output node.
placeholder(target, args, kwargs)	Execute a placeholder node.
run(*args[, initial_env, enable_io_processing])	Run module via interpretation and return the result.
run_node(n)	Run a specific node n and return the result.
transform()	Transform self.module and return the transformed GraphModule.