MCPcopy Create free account
hub / github.com/OpenGVLab/EfficientQAT / truncate_number

Function truncate_number

quantize/utils.py:95–97  ·  view source on GitHub ↗
(number, threshold=1e-2)

Source from the content-addressed store, hash-verified

93
94
95def truncate_number(number, threshold=1e-2):
96 # avoid overflow with AMP training
97 return TruncateFunction.apply(number, threshold)
98
99
100def get_named_linears(module, type):

Callers

nothing calls this directly

Calls

no outgoing calls

Tested by

no test coverage detected