fairseq/model_parallel/models/transformer_lm.py

Killed 4 out of 14 mutants

Survived

Survived mutation testing. These mutants show holes in your test suite.

Mutant 191

--- fairseq/model_parallel/models/transformer_lm.py
+++ fairseq/model_parallel/models/transformer_lm.py
@@ -15,7 +15,7 @@
     from fairseq.model_parallel.megatron.mpu import VocabParallelEmbedding
     has_megatron_submodule = True
 except (ImportError, ModuleNotFoundError):
-    has_megatron_submodule = False
+    has_megatron_submodule = True
 
 
 DEFAULT_MAX_TARGET_POSITIONS = 1024

Mutant 192

--- fairseq/model_parallel/models/transformer_lm.py
+++ fairseq/model_parallel/models/transformer_lm.py
@@ -15,7 +15,7 @@
     from fairseq.model_parallel.megatron.mpu import VocabParallelEmbedding
     has_megatron_submodule = True
 except (ImportError, ModuleNotFoundError):
-    has_megatron_submodule = False
+    has_megatron_submodule = None
 
 
 DEFAULT_MAX_TARGET_POSITIONS = 1024

Mutant 193

--- fairseq/model_parallel/models/transformer_lm.py
+++ fairseq/model_parallel/models/transformer_lm.py
@@ -18,7 +18,7 @@
     has_megatron_submodule = False
 
 
-DEFAULT_MAX_TARGET_POSITIONS = 1024
+DEFAULT_MAX_TARGET_POSITIONS = 1025
 
 
 @register_model('model_parallel_transformer_lm')

Mutant 194

--- fairseq/model_parallel/models/transformer_lm.py
+++ fairseq/model_parallel/models/transformer_lm.py
@@ -18,7 +18,7 @@
     has_megatron_submodule = False
 
 
-DEFAULT_MAX_TARGET_POSITIONS = 1024
+DEFAULT_MAX_TARGET_POSITIONS = None
 
 
 @register_model('model_parallel_transformer_lm')

Mutant 197

--- fairseq/model_parallel/models/transformer_lm.py
+++ fairseq/model_parallel/models/transformer_lm.py
@@ -24,7 +24,6 @@
 @register_model('model_parallel_transformer_lm')
 class ModelParallelTransformerLanguageModel(TransformerLanguageModel):
 
-    @classmethod
     def build_model(cls, args, task):
         """Build a new model instance."""
         if not has_megatron_submodule:

Mutant 198

--- fairseq/model_parallel/models/transformer_lm.py
+++ fairseq/model_parallel/models/transformer_lm.py
@@ -55,7 +55,6 @@
         )
         return cls(decoder)
 
-    @classmethod
     def build_embedding(cls, args, dictionary, embed_dim, path=None):
         def _vocab_init(tensor, **kwargs):
             nn.init.normal_(tensor, mean=0, std=embed_dim ** -0.5)

Mutant 200

--- fairseq/model_parallel/models/transformer_lm.py
+++ fairseq/model_parallel/models/transformer_lm.py
@@ -64,7 +64,7 @@
         return embed_tokens
 
 
-@register_model_architecture('model_parallel_transformer_lm', 'transformer_lm_megatron')
+@register_model_architecture('model_parallel_transformer_lm', 'XXtransformer_lm_megatronXX')
 def transformer_lm_megatron(args):
     args.decoder_embed_dim = getattr(args, 'decoder_embed_dim', 3072)
     args.decoder_ffn_embed_dim = getattr(args, 'decoder_ffn_embed_dim', 3072 * 4)

Mutant 201

--- fairseq/model_parallel/models/transformer_lm.py
+++ fairseq/model_parallel/models/transformer_lm.py
@@ -63,8 +63,6 @@
         embed_tokens = VocabParallelEmbedding(len(dictionary), embed_dim, dictionary.pad(), init_method=_vocab_init)
         return embed_tokens
 
-
-@register_model_architecture('model_parallel_transformer_lm', 'transformer_lm_megatron')
 def transformer_lm_megatron(args):
     args.decoder_embed_dim = getattr(args, 'decoder_embed_dim', 3072)
     args.decoder_ffn_embed_dim = getattr(args, 'decoder_ffn_embed_dim', 3072 * 4)

Mutant 203

--- fairseq/model_parallel/models/transformer_lm.py
+++ fairseq/model_parallel/models/transformer_lm.py
@@ -76,7 +76,7 @@
     base_lm_architecture(args)
 
 
-@register_model_architecture('model_parallel_transformer_lm', 'transformer_lm_megatron_11b')
+@register_model_architecture('model_parallel_transformer_lm', 'XXtransformer_lm_megatron_11bXX')
 def transformer_lm_megatron_11b(args):
     args.decoder_embed_dim = getattr(args, 'decoder_embed_dim', 3072)
     args.decoder_ffn_embed_dim = getattr(args, 'decoder_ffn_embed_dim', 3072 * 6)

Mutant 204

--- fairseq/model_parallel/models/transformer_lm.py
+++ fairseq/model_parallel/models/transformer_lm.py
@@ -75,8 +75,6 @@
     args.activation_fn = getattr(args, 'activation_fn', 'gelu')
     base_lm_architecture(args)
 
-
-@register_model_architecture('model_parallel_transformer_lm', 'transformer_lm_megatron_11b')
 def transformer_lm_megatron_11b(args):
     args.decoder_embed_dim = getattr(args, 'decoder_embed_dim', 3072)
     args.decoder_ffn_embed_dim = getattr(args, 'decoder_ffn_embed_dim', 3072 * 6)