fairseq/models/fconv_self_att.py
Killed 4 out of 71 mutantsSurvived
Survived mutation testing. These mutants show holes in your test suite.Mutant 1163
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -30,7 +30,7 @@
)
from fairseq.incremental_decoding_utils import with_incremental_state
-logger = logging.getLogger(__name__)
+logger = None
@register_model('fconv_self_att')
Mutant 1166
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -36,7 +36,6 @@
@register_model('fconv_self_att')
class FConvModelSelfAtt(FairseqEncoderDecoderModel):
- @classmethod
def hub_models(cls):
return {
'conv.stories.pretrained': {
Mutant 1167
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -67,7 +67,6 @@
# these are forwarded and then combined in the decoder
self.encoder = CompositeEncoder(encoders)
- @staticmethod
def add_args(parser):
"""Add model-specific arguments to the parser."""
# fmt: off
Mutant 1168
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -71,7 +71,7 @@
def add_args(parser):
"""Add model-specific arguments to the parser."""
# fmt: off
- parser.add_argument('--dropout', type=float, metavar='D',
+ parser.add_argument('XX--dropoutXX', type=float, metavar='D',
help='dropout probability')
parser.add_argument('--encoder-embed-dim', type=int, metavar='N',
help='encoder embedding dimension')
Mutant 1169
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -71,7 +71,7 @@
def add_args(parser):
"""Add model-specific arguments to the parser."""
# fmt: off
- parser.add_argument('--dropout', type=float, metavar='D',
+ parser.add_argument('--dropout', type=float, metavar='XXDXX',
help='dropout probability')
parser.add_argument('--encoder-embed-dim', type=int, metavar='N',
help='encoder embedding dimension')
Mutant 1170
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -72,7 +72,7 @@
"""Add model-specific arguments to the parser."""
# fmt: off
parser.add_argument('--dropout', type=float, metavar='D',
- help='dropout probability')
+ help='XXdropout probabilityXX')
parser.add_argument('--encoder-embed-dim', type=int, metavar='N',
help='encoder embedding dimension')
parser.add_argument('--encoder-layers', type=str, metavar='EXPR',
Mutant 1171
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -73,7 +73,7 @@
# fmt: off
parser.add_argument('--dropout', type=float, metavar='D',
help='dropout probability')
- parser.add_argument('--encoder-embed-dim', type=int, metavar='N',
+ parser.add_argument('XX--encoder-embed-dimXX', type=int, metavar='N',
help='encoder embedding dimension')
parser.add_argument('--encoder-layers', type=str, metavar='EXPR',
help='encoder layers [(dim, kernel_size), ...]')
Mutant 1172
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -73,7 +73,7 @@
# fmt: off
parser.add_argument('--dropout', type=float, metavar='D',
help='dropout probability')
- parser.add_argument('--encoder-embed-dim', type=int, metavar='N',
+ parser.add_argument('--encoder-embed-dim', type=int, metavar='XXNXX',
help='encoder embedding dimension')
parser.add_argument('--encoder-layers', type=str, metavar='EXPR',
help='encoder layers [(dim, kernel_size), ...]')
Mutant 1173
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -74,7 +74,7 @@
parser.add_argument('--dropout', type=float, metavar='D',
help='dropout probability')
parser.add_argument('--encoder-embed-dim', type=int, metavar='N',
- help='encoder embedding dimension')
+ help='XXencoder embedding dimensionXX')
parser.add_argument('--encoder-layers', type=str, metavar='EXPR',
help='encoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-embed-dim', type=int, metavar='N',
Mutant 1174
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -75,7 +75,7 @@
help='dropout probability')
parser.add_argument('--encoder-embed-dim', type=int, metavar='N',
help='encoder embedding dimension')
- parser.add_argument('--encoder-layers', type=str, metavar='EXPR',
+ parser.add_argument('XX--encoder-layersXX', type=str, metavar='EXPR',
help='encoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-embed-dim', type=int, metavar='N',
help='decoder embedding dimension')
Mutant 1175
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -75,7 +75,7 @@
help='dropout probability')
parser.add_argument('--encoder-embed-dim', type=int, metavar='N',
help='encoder embedding dimension')
- parser.add_argument('--encoder-layers', type=str, metavar='EXPR',
+ parser.add_argument('--encoder-layers', type=str, metavar='XXEXPRXX',
help='encoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-embed-dim', type=int, metavar='N',
help='decoder embedding dimension')
Mutant 1176
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -76,7 +76,7 @@
parser.add_argument('--encoder-embed-dim', type=int, metavar='N',
help='encoder embedding dimension')
parser.add_argument('--encoder-layers', type=str, metavar='EXPR',
- help='encoder layers [(dim, kernel_size), ...]')
+ help='XXencoder layers [(dim, kernel_size), ...]XX')
parser.add_argument('--decoder-embed-dim', type=int, metavar='N',
help='decoder embedding dimension')
parser.add_argument('--decoder-layers', type=str, metavar='EXPR',
Mutant 1177
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -77,7 +77,7 @@
help='encoder embedding dimension')
parser.add_argument('--encoder-layers', type=str, metavar='EXPR',
help='encoder layers [(dim, kernel_size), ...]')
- parser.add_argument('--decoder-embed-dim', type=int, metavar='N',
+ parser.add_argument('XX--decoder-embed-dimXX', type=int, metavar='N',
help='decoder embedding dimension')
parser.add_argument('--decoder-layers', type=str, metavar='EXPR',
help='decoder layers [(dim, kernel_size), ...]')
Mutant 1178
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -77,7 +77,7 @@
help='encoder embedding dimension')
parser.add_argument('--encoder-layers', type=str, metavar='EXPR',
help='encoder layers [(dim, kernel_size), ...]')
- parser.add_argument('--decoder-embed-dim', type=int, metavar='N',
+ parser.add_argument('--decoder-embed-dim', type=int, metavar='XXNXX',
help='decoder embedding dimension')
parser.add_argument('--decoder-layers', type=str, metavar='EXPR',
help='decoder layers [(dim, kernel_size), ...]')
Mutant 1179
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -78,7 +78,7 @@
parser.add_argument('--encoder-layers', type=str, metavar='EXPR',
help='encoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-embed-dim', type=int, metavar='N',
- help='decoder embedding dimension')
+ help='XXdecoder embedding dimensionXX')
parser.add_argument('--decoder-layers', type=str, metavar='EXPR',
help='decoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-out-embed-dim', type=int, metavar='N',
Mutant 1180
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -79,7 +79,7 @@
help='encoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-embed-dim', type=int, metavar='N',
help='decoder embedding dimension')
- parser.add_argument('--decoder-layers', type=str, metavar='EXPR',
+ parser.add_argument('XX--decoder-layersXX', type=str, metavar='EXPR',
help='decoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-out-embed-dim', type=int, metavar='N',
help='decoder output embedding dimension')
Mutant 1181
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -79,7 +79,7 @@
help='encoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-embed-dim', type=int, metavar='N',
help='decoder embedding dimension')
- parser.add_argument('--decoder-layers', type=str, metavar='EXPR',
+ parser.add_argument('--decoder-layers', type=str, metavar='XXEXPRXX',
help='decoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-out-embed-dim', type=int, metavar='N',
help='decoder output embedding dimension')
Mutant 1182
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -80,7 +80,7 @@
parser.add_argument('--decoder-embed-dim', type=int, metavar='N',
help='decoder embedding dimension')
parser.add_argument('--decoder-layers', type=str, metavar='EXPR',
- help='decoder layers [(dim, kernel_size), ...]')
+ help='XXdecoder layers [(dim, kernel_size), ...]XX')
parser.add_argument('--decoder-out-embed-dim', type=int, metavar='N',
help='decoder output embedding dimension')
parser.add_argument('--decoder-attention', type=str, metavar='EXPR',
Mutant 1183
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -81,7 +81,7 @@
help='decoder embedding dimension')
parser.add_argument('--decoder-layers', type=str, metavar='EXPR',
help='decoder layers [(dim, kernel_size), ...]')
- parser.add_argument('--decoder-out-embed-dim', type=int, metavar='N',
+ parser.add_argument('XX--decoder-out-embed-dimXX', type=int, metavar='N',
help='decoder output embedding dimension')
parser.add_argument('--decoder-attention', type=str, metavar='EXPR',
help='decoder attention [True, ...]')
Mutant 1184
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -81,7 +81,7 @@
help='decoder embedding dimension')
parser.add_argument('--decoder-layers', type=str, metavar='EXPR',
help='decoder layers [(dim, kernel_size), ...]')
- parser.add_argument('--decoder-out-embed-dim', type=int, metavar='N',
+ parser.add_argument('--decoder-out-embed-dim', type=int, metavar='XXNXX',
help='decoder output embedding dimension')
parser.add_argument('--decoder-attention', type=str, metavar='EXPR',
help='decoder attention [True, ...]')
Mutant 1185
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -82,7 +82,7 @@
parser.add_argument('--decoder-layers', type=str, metavar='EXPR',
help='decoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-out-embed-dim', type=int, metavar='N',
- help='decoder output embedding dimension')
+ help='XXdecoder output embedding dimensionXX')
parser.add_argument('--decoder-attention', type=str, metavar='EXPR',
help='decoder attention [True, ...]')
parser.add_argument('--self-attention', type=str, metavar='EXPR',
Mutant 1186
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -83,7 +83,7 @@
help='decoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-out-embed-dim', type=int, metavar='N',
help='decoder output embedding dimension')
- parser.add_argument('--decoder-attention', type=str, metavar='EXPR',
+ parser.add_argument('XX--decoder-attentionXX', type=str, metavar='EXPR',
help='decoder attention [True, ...]')
parser.add_argument('--self-attention', type=str, metavar='EXPR',
help='decoder self-attention layers, ex: [True] + [False]*5')
Mutant 1187
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -83,7 +83,7 @@
help='decoder layers [(dim, kernel_size), ...]')
parser.add_argument('--decoder-out-embed-dim', type=int, metavar='N',
help='decoder output embedding dimension')
- parser.add_argument('--decoder-attention', type=str, metavar='EXPR',
+ parser.add_argument('--decoder-attention', type=str, metavar='XXEXPRXX',
help='decoder attention [True, ...]')
parser.add_argument('--self-attention', type=str, metavar='EXPR',
help='decoder self-attention layers, ex: [True] + [False]*5')
Mutant 1188
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -84,7 +84,7 @@
parser.add_argument('--decoder-out-embed-dim', type=int, metavar='N',
help='decoder output embedding dimension')
parser.add_argument('--decoder-attention', type=str, metavar='EXPR',
- help='decoder attention [True, ...]')
+ help='XXdecoder attention [True, ...]XX')
parser.add_argument('--self-attention', type=str, metavar='EXPR',
help='decoder self-attention layers, ex: [True] + [False]*5')
parser.add_argument('--multihead-attention-nheads', type=int,
Mutant 1189
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -85,7 +85,7 @@
help='decoder output embedding dimension')
parser.add_argument('--decoder-attention', type=str, metavar='EXPR',
help='decoder attention [True, ...]')
- parser.add_argument('--self-attention', type=str, metavar='EXPR',
+ parser.add_argument('XX--self-attentionXX', type=str, metavar='EXPR',
help='decoder self-attention layers, ex: [True] + [False]*5')
parser.add_argument('--multihead-attention-nheads', type=int,
help='Number of heads to use in attention')
Mutant 1190
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -85,7 +85,7 @@
help='decoder output embedding dimension')
parser.add_argument('--decoder-attention', type=str, metavar='EXPR',
help='decoder attention [True, ...]')
- parser.add_argument('--self-attention', type=str, metavar='EXPR',
+ parser.add_argument('--self-attention', type=str, metavar='XXEXPRXX',
help='decoder self-attention layers, ex: [True] + [False]*5')
parser.add_argument('--multihead-attention-nheads', type=int,
help='Number of heads to use in attention')
Mutant 1191
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -86,7 +86,7 @@
parser.add_argument('--decoder-attention', type=str, metavar='EXPR',
help='decoder attention [True, ...]')
parser.add_argument('--self-attention', type=str, metavar='EXPR',
- help='decoder self-attention layers, ex: [True] + [False]*5')
+ help='XXdecoder self-attention layers, ex: [True] + [False]*5XX')
parser.add_argument('--multihead-attention-nheads', type=int,
help='Number of heads to use in attention')
parser.add_argument('--multihead-self-attention-nheads', type=int,
Mutant 1192
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -87,7 +87,7 @@
help='decoder attention [True, ...]')
parser.add_argument('--self-attention', type=str, metavar='EXPR',
help='decoder self-attention layers, ex: [True] + [False]*5')
- parser.add_argument('--multihead-attention-nheads', type=int,
+ parser.add_argument('XX--multihead-attention-nheadsXX', type=int,
help='Number of heads to use in attention')
parser.add_argument('--multihead-self-attention-nheads', type=int,
help='Number of heads to use in self-attention')
Mutant 1193
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -88,7 +88,7 @@
parser.add_argument('--self-attention', type=str, metavar='EXPR',
help='decoder self-attention layers, ex: [True] + [False]*5')
parser.add_argument('--multihead-attention-nheads', type=int,
- help='Number of heads to use in attention')
+ help='XXNumber of heads to use in attentionXX')
parser.add_argument('--multihead-self-attention-nheads', type=int,
help='Number of heads to use in self-attention')
parser.add_argument('--encoder-attention', type=str, metavar='EXPR',
Mutant 1194
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -89,7 +89,7 @@
help='decoder self-attention layers, ex: [True] + [False]*5')
parser.add_argument('--multihead-attention-nheads', type=int,
help='Number of heads to use in attention')
- parser.add_argument('--multihead-self-attention-nheads', type=int,
+ parser.add_argument('XX--multihead-self-attention-nheadsXX', type=int,
help='Number of heads to use in self-attention')
parser.add_argument('--encoder-attention', type=str, metavar='EXPR',
help='encoder attention [True, ...]')
Mutant 1195
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -90,7 +90,7 @@
parser.add_argument('--multihead-attention-nheads', type=int,
help='Number of heads to use in attention')
parser.add_argument('--multihead-self-attention-nheads', type=int,
- help='Number of heads to use in self-attention')
+ help='XXNumber of heads to use in self-attentionXX')
parser.add_argument('--encoder-attention', type=str, metavar='EXPR',
help='encoder attention [True, ...]')
parser.add_argument('--encoder-attention-nheads', type=int,
Mutant 1196
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -91,7 +91,7 @@
help='Number of heads to use in attention')
parser.add_argument('--multihead-self-attention-nheads', type=int,
help='Number of heads to use in self-attention')
- parser.add_argument('--encoder-attention', type=str, metavar='EXPR',
+ parser.add_argument('XX--encoder-attentionXX', type=str, metavar='EXPR',
help='encoder attention [True, ...]')
parser.add_argument('--encoder-attention-nheads', type=int,
help='Number of heads to use in encoder attention')
Mutant 1197
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -91,7 +91,7 @@
help='Number of heads to use in attention')
parser.add_argument('--multihead-self-attention-nheads', type=int,
help='Number of heads to use in self-attention')
- parser.add_argument('--encoder-attention', type=str, metavar='EXPR',
+ parser.add_argument('--encoder-attention', type=str, metavar='XXEXPRXX',
help='encoder attention [True, ...]')
parser.add_argument('--encoder-attention-nheads', type=int,
help='Number of heads to use in encoder attention')
Mutant 1198
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -92,7 +92,7 @@
parser.add_argument('--multihead-self-attention-nheads', type=int,
help='Number of heads to use in self-attention')
parser.add_argument('--encoder-attention', type=str, metavar='EXPR',
- help='encoder attention [True, ...]')
+ help='XXencoder attention [True, ...]XX')
parser.add_argument('--encoder-attention-nheads', type=int,
help='Number of heads to use in encoder attention')
parser.add_argument('--project-input', type=str, metavar='EXPR',
Mutant 1199
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -93,7 +93,7 @@
help='Number of heads to use in self-attention')
parser.add_argument('--encoder-attention', type=str, metavar='EXPR',
help='encoder attention [True, ...]')
- parser.add_argument('--encoder-attention-nheads', type=int,
+ parser.add_argument('XX--encoder-attention-nheadsXX', type=int,
help='Number of heads to use in encoder attention')
parser.add_argument('--project-input', type=str, metavar='EXPR',
help='Use projections in self-attention [True, ...]')
Mutant 1200
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -94,7 +94,7 @@
parser.add_argument('--encoder-attention', type=str, metavar='EXPR',
help='encoder attention [True, ...]')
parser.add_argument('--encoder-attention-nheads', type=int,
- help='Number of heads to use in encoder attention')
+ help='XXNumber of heads to use in encoder attentionXX')
parser.add_argument('--project-input', type=str, metavar='EXPR',
help='Use projections in self-attention [True, ...]')
parser.add_argument('--gated-attention', type=str, metavar='EXPR',
Mutant 1201
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -95,7 +95,7 @@
help='encoder attention [True, ...]')
parser.add_argument('--encoder-attention-nheads', type=int,
help='Number of heads to use in encoder attention')
- parser.add_argument('--project-input', type=str, metavar='EXPR',
+ parser.add_argument('XX--project-inputXX', type=str, metavar='EXPR',
help='Use projections in self-attention [True, ...]')
parser.add_argument('--gated-attention', type=str, metavar='EXPR',
help='Use GLU layers in self-attention projections [True, ...]')
Mutant 1202
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -95,7 +95,7 @@
help='encoder attention [True, ...]')
parser.add_argument('--encoder-attention-nheads', type=int,
help='Number of heads to use in encoder attention')
- parser.add_argument('--project-input', type=str, metavar='EXPR',
+ parser.add_argument('--project-input', type=str, metavar='XXEXPRXX',
help='Use projections in self-attention [True, ...]')
parser.add_argument('--gated-attention', type=str, metavar='EXPR',
help='Use GLU layers in self-attention projections [True, ...]')
Mutant 1203
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -96,7 +96,7 @@
parser.add_argument('--encoder-attention-nheads', type=int,
help='Number of heads to use in encoder attention')
parser.add_argument('--project-input', type=str, metavar='EXPR',
- help='Use projections in self-attention [True, ...]')
+ help='XXUse projections in self-attention [True, ...]XX')
parser.add_argument('--gated-attention', type=str, metavar='EXPR',
help='Use GLU layers in self-attention projections [True, ...]')
parser.add_argument('--downsample', type=str, metavar='EXPR',
Mutant 1204
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -97,7 +97,7 @@
help='Number of heads to use in encoder attention')
parser.add_argument('--project-input', type=str, metavar='EXPR',
help='Use projections in self-attention [True, ...]')
- parser.add_argument('--gated-attention', type=str, metavar='EXPR',
+ parser.add_argument('XX--gated-attentionXX', type=str, metavar='EXPR',
help='Use GLU layers in self-attention projections [True, ...]')
parser.add_argument('--downsample', type=str, metavar='EXPR',
help='Use downsampling in self-attention [True, ...]')
Mutant 1205
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -97,7 +97,7 @@
help='Number of heads to use in encoder attention')
parser.add_argument('--project-input', type=str, metavar='EXPR',
help='Use projections in self-attention [True, ...]')
- parser.add_argument('--gated-attention', type=str, metavar='EXPR',
+ parser.add_argument('--gated-attention', type=str, metavar='XXEXPRXX',
help='Use GLU layers in self-attention projections [True, ...]')
parser.add_argument('--downsample', type=str, metavar='EXPR',
help='Use downsampling in self-attention [True, ...]')
Mutant 1206
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -98,7 +98,7 @@
parser.add_argument('--project-input', type=str, metavar='EXPR',
help='Use projections in self-attention [True, ...]')
parser.add_argument('--gated-attention', type=str, metavar='EXPR',
- help='Use GLU layers in self-attention projections [True, ...]')
+ help='XXUse GLU layers in self-attention projections [True, ...]XX')
parser.add_argument('--downsample', type=str, metavar='EXPR',
help='Use downsampling in self-attention [True, ...]')
parser.add_argument('--pretrained-checkpoint', metavar='DIR',
Mutant 1207
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -99,7 +99,7 @@
help='Use projections in self-attention [True, ...]')
parser.add_argument('--gated-attention', type=str, metavar='EXPR',
help='Use GLU layers in self-attention projections [True, ...]')
- parser.add_argument('--downsample', type=str, metavar='EXPR',
+ parser.add_argument('XX--downsampleXX', type=str, metavar='EXPR',
help='Use downsampling in self-attention [True, ...]')
parser.add_argument('--pretrained-checkpoint', metavar='DIR',
help='path to load checkpoint from pretrained model')
Mutant 1208
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -99,7 +99,7 @@
help='Use projections in self-attention [True, ...]')
parser.add_argument('--gated-attention', type=str, metavar='EXPR',
help='Use GLU layers in self-attention projections [True, ...]')
- parser.add_argument('--downsample', type=str, metavar='EXPR',
+ parser.add_argument('--downsample', type=str, metavar='XXEXPRXX',
help='Use downsampling in self-attention [True, ...]')
parser.add_argument('--pretrained-checkpoint', metavar='DIR',
help='path to load checkpoint from pretrained model')
Mutant 1209
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -100,7 +100,7 @@
parser.add_argument('--gated-attention', type=str, metavar='EXPR',
help='Use GLU layers in self-attention projections [True, ...]')
parser.add_argument('--downsample', type=str, metavar='EXPR',
- help='Use downsampling in self-attention [True, ...]')
+ help='XXUse downsampling in self-attention [True, ...]XX')
parser.add_argument('--pretrained-checkpoint', metavar='DIR',
help='path to load checkpoint from pretrained model')
parser.add_argument('--pretrained', type=str, metavar='EXPR',
Mutant 1210
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -101,7 +101,7 @@
help='Use GLU layers in self-attention projections [True, ...]')
parser.add_argument('--downsample', type=str, metavar='EXPR',
help='Use downsampling in self-attention [True, ...]')
- parser.add_argument('--pretrained-checkpoint', metavar='DIR',
+ parser.add_argument('XX--pretrained-checkpointXX', metavar='DIR',
help='path to load checkpoint from pretrained model')
parser.add_argument('--pretrained', type=str, metavar='EXPR',
help='use pretrained model when training [True, ...]')
Mutant 1211
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -101,7 +101,7 @@
help='Use GLU layers in self-attention projections [True, ...]')
parser.add_argument('--downsample', type=str, metavar='EXPR',
help='Use downsampling in self-attention [True, ...]')
- parser.add_argument('--pretrained-checkpoint', metavar='DIR',
+ parser.add_argument('--pretrained-checkpoint', metavar='XXDIRXX',
help='path to load checkpoint from pretrained model')
parser.add_argument('--pretrained', type=str, metavar='EXPR',
help='use pretrained model when training [True, ...]')
Mutant 1212
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -102,7 +102,7 @@
parser.add_argument('--downsample', type=str, metavar='EXPR',
help='Use downsampling in self-attention [True, ...]')
parser.add_argument('--pretrained-checkpoint', metavar='DIR',
- help='path to load checkpoint from pretrained model')
+ help='XXpath to load checkpoint from pretrained modelXX')
parser.add_argument('--pretrained', type=str, metavar='EXPR',
help='use pretrained model when training [True, ...]')
# fmt: on
Mutant 1213
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -103,7 +103,7 @@
help='Use downsampling in self-attention [True, ...]')
parser.add_argument('--pretrained-checkpoint', metavar='DIR',
help='path to load checkpoint from pretrained model')
- parser.add_argument('--pretrained', type=str, metavar='EXPR',
+ parser.add_argument('XX--pretrainedXX', type=str, metavar='EXPR',
help='use pretrained model when training [True, ...]')
# fmt: on
Mutant 1214
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -103,7 +103,7 @@
help='Use downsampling in self-attention [True, ...]')
parser.add_argument('--pretrained-checkpoint', metavar='DIR',
help='path to load checkpoint from pretrained model')
- parser.add_argument('--pretrained', type=str, metavar='EXPR',
+ parser.add_argument('--pretrained', type=str, metavar='XXEXPRXX',
help='use pretrained model when training [True, ...]')
# fmt: on
Mutant 1215
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -104,7 +104,7 @@
parser.add_argument('--pretrained-checkpoint', metavar='DIR',
help='path to load checkpoint from pretrained model')
parser.add_argument('--pretrained', type=str, metavar='EXPR',
- help='use pretrained model when training [True, ...]')
+ help='XXuse pretrained model when training [True, ...]XX')
# fmt: on
@classmethod
Mutant 1216
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -105,9 +105,7 @@
help='path to load checkpoint from pretrained model')
parser.add_argument('--pretrained', type=str, metavar='EXPR',
help='use pretrained model when training [True, ...]')
- # fmt: on
-
- @classmethod
+
def build_model(cls, args, task):
"""Build a new model instance."""
trained_encoder, trained_decoder = None, None
Mutant 1217
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -162,7 +162,6 @@
return model
- @property
def pretrained(self):
return self.pretrained_encoder is not None
Mutant 1218
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -172,7 +172,7 @@
def __init__(
self, dictionary, embed_dim=512, max_positions=1024,
convolutions=((512, 3),) * 20, dropout=0.1, attention=False,
- attention_nheads=1,
+ attention_nheads=2,
):
super().__init__(dictionary)
self.dropout_module = FairseqDropout(
Mutant 1219
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -293,8 +293,6 @@
"""Maximum input length supported by the encoder."""
return self.embed_positions.max_positions
-
-@with_incremental_state
class FConvDecoder(FairseqDecoder):
"""Convolutional decoder"""
def __init__(
Mutant 1220
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -302,7 +302,7 @@
convolutions=((512, 3),) * 8, attention=True, dropout=0.1,
selfattention=False, attention_nheads=1, selfattention_nheads=1,
project_input=False, gated_attention=False, downsample=False,
- pretrained=False, trained_decoder=None,
+ pretrained=True, trained_decoder=None,
):
super().__init__(dictionary)
self.register_buffer('version', torch.Tensor([2]))
Mutant 1221
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -479,7 +479,7 @@
"""Maximum output length supported by the decoder."""
return self.embed_positions.max_positions
- def make_generation_fast_(self, need_attn=False, **kwargs):
+ def make_generation_fast_(self, need_attn=True, **kwargs):
self.need_attn = need_attn
def _split_encoder_out(self, encoder_out):
Mutant 1222
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -494,7 +494,7 @@
class SelfAttention(nn.Module):
- def __init__(self, out_channels, embed_dim, num_heads, project_input=False, gated=False, downsample=False):
+ def __init__(self, out_channels, embed_dim, num_heads, project_input=True, gated=False, downsample=False):
super().__init__()
self.attention = DownsampledMultiHeadAttention(
out_channels, embed_dim, num_heads, dropout=0, bias=True,
Mutant 1223
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -494,7 +494,7 @@
class SelfAttention(nn.Module):
- def __init__(self, out_channels, embed_dim, num_heads, project_input=False, gated=False, downsample=False):
+ def __init__(self, out_channels, embed_dim, num_heads, project_input=False, gated=True, downsample=False):
super().__init__()
self.attention = DownsampledMultiHeadAttention(
out_channels, embed_dim, num_heads, dropout=0, bias=True,
Mutant 1224
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -494,7 +494,7 @@
class SelfAttention(nn.Module):
- def __init__(self, out_channels, embed_dim, num_heads, project_input=False, gated=False, downsample=False):
+ def __init__(self, out_channels, embed_dim, num_heads, project_input=False, gated=False, downsample=True):
super().__init__()
self.attention = DownsampledMultiHeadAttention(
out_channels, embed_dim, num_heads, dropout=0, bias=True,
Mutant 1225
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -526,7 +526,7 @@
return m
-def Linear(in_features, out_features, dropout=0.):
+def Linear(in_features, out_features, dropout=1.0):
"""Weight-normalized Linear layer (input: N x T x C)"""
m = nn.Linear(in_features, out_features)
m.weight.data.normal_(mean=0, std=math.sqrt((1 - dropout) / in_features))
Mutant 1226
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -534,7 +534,7 @@
return m
-def LinearizedConv1d(in_channels, out_channels, kernel_size, dropout=0., **kwargs):
+def LinearizedConv1d(in_channels, out_channels, kernel_size, dropout=1.0, **kwargs):
"""Weight-normalized Conv1d layer optimized for decoding"""
m = LinearizedConvolution(in_channels, out_channels, kernel_size, **kwargs)
std = math.sqrt((4 * (1.0 - dropout)) / (m.kernel_size[0] * in_channels))
Mutant 1227
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -543,7 +543,7 @@
return m
-def ConvTBC(in_channels, out_channels, kernel_size, dropout=0., **kwargs):
+def ConvTBC(in_channels, out_channels, kernel_size, dropout=1.0, **kwargs):
"""Weight-normalized Conv1d layer"""
from fairseq.modules import ConvTBC
m = ConvTBC(in_channels, out_channels, kernel_size, **kwargs)
Mutant 1229
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -553,7 +553,7 @@
return m
-@register_model_architecture('fconv_self_att', 'fconv_self_att')
+@register_model_architecture('fconv_self_att', 'XXfconv_self_attXX')
def base_architecture(args):
args.dropout = getattr(args, 'dropout', 0.1)
args.encoder_embed_dim = getattr(args, 'encoder_embed_dim', 512)
Mutant 1230
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -552,8 +552,6 @@
m.bias.data.zero_()
return m
-
-@register_model_architecture('fconv_self_att', 'fconv_self_att')
def base_architecture(args):
args.dropout = getattr(args, 'dropout', 0.1)
args.encoder_embed_dim = getattr(args, 'encoder_embed_dim', 512)
Mutant 1232
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -574,7 +574,7 @@
args.pretrained = getattr(args, 'pretrained', 'False')
-@register_model_architecture('fconv_self_att', 'fconv_self_att_wp')
+@register_model_architecture('fconv_self_att', 'XXfconv_self_att_wpXX')
def fconv_self_att_wp(args):
args.encoder_embed_dim = getattr(args, 'encoder_embed_dim', 256)
args.encoder_layers = getattr(args, 'encoder_layers', '[(128, 3)] * 2 + [(512,3)] * 1')
Mutant 1233
--- fairseq/models/fconv_self_att.py
+++ fairseq/models/fconv_self_att.py
@@ -573,8 +573,6 @@
args.pretrained_checkpoint = getattr(args, 'pretrained_checkpoint', '')
args.pretrained = getattr(args, 'pretrained', 'False')
-
-@register_model_architecture('fconv_self_att', 'fconv_self_att_wp')
def fconv_self_att_wp(args):
args.encoder_embed_dim = getattr(args, 'encoder_embed_dim', 256)
args.encoder_layers = getattr(args, 'encoder_layers', '[(128, 3)] * 2 + [(512,3)] * 1')