OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-06 12:20:52 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity

migrate mtml to dper2

Browse Source 
Summary:
1. migrate the basic mtml model to dper 2
2. test dper 2 mtml model
3. test all optimizers

Reviewed By: kittipatv

Differential Revision: D4680215

fbshipit-source-id: 7aac5c59bdac22fcad8ed869b98e9e62dca1d337
This commit is contained in:
Huazhong Ning 2017-03-16 17:43:48 -07:00 committed by Facebook Github Bot
parent cc2e915461
commit ad4ae4528f
2 changed files with 24 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
caffe2/python/layers/batch_lr_loss.py
View File

@ -3,7 +3,7 @@ from __future__ import division
from __future__ import print_function
from __future__ import unicode_literals
from caffe2.python import core, schema
from caffe2.python import schema
from caffe2.python.layers.layers import (
ModelLayer,
)
@ -15,10 +15,13 @@ import numpy as np
class BatchLRLoss(ModelLayer):
def __init__(self, model, input_record, name='batch_lr_loss', **kwargs):
def __init__(self, model, input_record, name='batch_lr_loss',
average_loss=True, **kwargs):
super(BatchLRLoss, self).__init__(model, name, input_record, **kwargs)
schema.is_schema_subset(
self.average_loss = average_loss
assert schema.is_schema_subset(
schema.Struct(
('label', schema.Scalar()),
('prediction', schema.Scalar())
@ -46,4 +49,13 @@ class BatchLRLoss(ModelLayer):
[class_probabilities] + label,
net.NextScopedBlob('cross_entropy'),
)
net.AveragedLoss(xent, self.output_schema.field_blobs())
if 'weight' in self.input_record.fields:
xent = net.Mul(
[xent, self.input_record.weight()],
net.NextScopedBlob('weighted_scross_entropy'),
)
if self.average_loss:
net.AveragedLoss(xent, self.output_schema.field_blobs())
else:
net.ReduceFrontSum(xent, self.output_schema.field_blobs())

11
caffe2/python/schema.py
View File

@ -635,20 +635,25 @@ def Map(
)
def NamedTuple(name_prefix, *fields):
return Struct(* [('%s_%d' % (name_prefix, i), field)
for i, field in enumerate(fields)])
def Tuple(*fields):
"""
Creates a Struct with default, sequential, field names of given types.
"""
return Struct(* [('field_%d' % i, field) for i, field in enumerate(fields)])
return NamedTuple('field', *fields)
def RawTuple(num_fields):
def RawTuple(num_fields, name_prefix='field'):
"""
Creates a tuple of `num_field` untyped scalars.
"""
assert isinstance(num_fields, int)
assert num_fields >= 0
return Tuple(*([np.void] * num_fields))
return NamedTuple(name_prefix, *([np.void] * num_fields))
def from_dtype(dtype, _outer_shape=()):