Lei Zhang b45069b59f fix fc fp16 quantization (#29469) ... Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/29469 The original approach is to save both fp16 and fp32 for all models, which increased the filesize and memory. This diff is to save 'used' blobs into predictor file. Test Plan: fc clone workflow : f149878151 ctr mbl feed test with fc fp16 quantization: f149996395 No fp32 in local file {F221750392} QRT after the fix: https://fburl.com/qrt/cp8r8263 Reviewed By: wx1988 Differential Revision: D18382503 fbshipit-source-id: 231c41668f25b1d35ca8d4358ce9b12ba60a4f91		2019-11-18 11:26:49 -08:00
..
__init__.py	Experimental support for setup.py develop mode install	2018-02-12 23:36:18 -08:00
mobile_exporter_test.py	Allow mobile exporter to export string arrays (#10017)	2018-08-01 16:09:58 -07:00
mobile_exporter.py	Caffe2: Fix for creating entries of external_input in predic_net (#12979)	2018-11-15 22:33:50 -08:00
predictor_exporter_test.py	fix fc fp16 quantization (#29469)	2019-11-18 11:26:49 -08:00
predictor_exporter.py	Supporting Manifold DB in Predictor Exporter (#22334)	2019-07-01 16:17:02 -07:00
predictor_py_utils.py	fix fc fp16 quantization (#29469)	2019-11-18 11:26:49 -08:00
predictor_test.py	Expose Predictor::run_map to Python	2017-10-18 19:32:56 -07:00
serde.py	Remove Apache headers from source.	2018-03-27 13:10:18 -07:00