fairseq/data/token_block_dataset.py
Killed 22 out of 39 mutantsSurvived
Survived mutation testing. These mutants show holes in your test suite.Mutant 2148
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -40,7 +40,7 @@
eos,
break_mode=None,
include_targets=False,
- document_sep_len=1,
+ document_sep_len=2,
):
try:
from fairseq.data.token_block_utils_fast import (
Mutant 2150
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -55,7 +55,7 @@
super().__init__()
self.dataset = dataset
- self.pad = pad
+ self.pad = None
self.eos = eos
self.include_targets = include_targets
Mutant 2151
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -56,7 +56,7 @@
super().__init__()
self.dataset = dataset
self.pad = pad
- self.eos = eos
+ self.eos = None
self.include_targets = include_targets
assert len(dataset) == len(sizes)
Mutant 2152
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -57,7 +57,7 @@
self.dataset = dataset
self.pad = pad
self.eos = eos
- self.include_targets = include_targets
+ self.include_targets = None
assert len(dataset) == len(sizes)
assert len(dataset) > 0
Mutant 2154
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -60,7 +60,7 @@
self.include_targets = include_targets
assert len(dataset) == len(sizes)
- assert len(dataset) > 0
+ assert len(dataset) >= 0
if isinstance(sizes, list):
sizes = np.array(sizes, dtype=np.int64)
Mutant 2157
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -69,7 +69,7 @@
sizes = sizes.numpy()
sizes = sizes.astype(np.int64)
- break_mode = break_mode if break_mode is not None else 'none'
+ break_mode = break_mode if break_mode is None else 'none'
# For "eos" break-mode, block_size is not required parameters.
if break_mode == "eos" and block_size is None:
Mutant 2159
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -69,7 +69,7 @@
sizes = sizes.numpy()
sizes = sizes.astype(np.int64)
- break_mode = break_mode if break_mode is not None else 'none'
+ break_mode = None
# For "eos" break-mode, block_size is not required parameters.
if break_mode == "eos" and block_size is None:
Mutant 2160
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -72,7 +72,7 @@
break_mode = break_mode if break_mode is not None else 'none'
# For "eos" break-mode, block_size is not required parameters.
- if break_mode == "eos" and block_size is None:
+ if break_mode != "eos" and block_size is None:
block_size = 0
slice_indices = _get_slice_indices_fast(sizes, break_mode, block_size, document_sep_len)
Mutant 2161
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -72,7 +72,7 @@
break_mode = break_mode if break_mode is not None else 'none'
# For "eos" break-mode, block_size is not required parameters.
- if break_mode == "eos" and block_size is None:
+ if break_mode == "XXeosXX" and block_size is None:
block_size = 0
slice_indices = _get_slice_indices_fast(sizes, break_mode, block_size, document_sep_len)
Mutant 2162
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -72,7 +72,7 @@
break_mode = break_mode if break_mode is not None else 'none'
# For "eos" break-mode, block_size is not required parameters.
- if break_mode == "eos" and block_size is None:
+ if break_mode == "eos" and block_size is not None:
block_size = 0
slice_indices = _get_slice_indices_fast(sizes, break_mode, block_size, document_sep_len)
Mutant 2163
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -72,7 +72,7 @@
break_mode = break_mode if break_mode is not None else 'none'
# For "eos" break-mode, block_size is not required parameters.
- if break_mode == "eos" and block_size is None:
+ if break_mode == "eos" or block_size is None:
block_size = 0
slice_indices = _get_slice_indices_fast(sizes, break_mode, block_size, document_sep_len)
Mutant 2166
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -76,7 +76,7 @@
block_size = 0
slice_indices = _get_slice_indices_fast(sizes, break_mode, block_size, document_sep_len)
- self._sizes = slice_indices[:, 1] - slice_indices[:, 0]
+ self._sizes = slice_indices[:, 1] + slice_indices[:, 0]
# build index mapping block indices to the underlying dataset indices
if break_mode == "eos":
Mutant 2167
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -76,7 +76,7 @@
block_size = 0
slice_indices = _get_slice_indices_fast(sizes, break_mode, block_size, document_sep_len)
- self._sizes = slice_indices[:, 1] - slice_indices[:, 0]
+ self._sizes = slice_indices[:, 1] - slice_indices[:, 1]
# build index mapping block indices to the underlying dataset indices
if break_mode == "eos":
Mutant 2169
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -79,7 +79,7 @@
self._sizes = slice_indices[:, 1] - slice_indices[:, 0]
# build index mapping block indices to the underlying dataset indices
- if break_mode == "eos":
+ if break_mode != "eos":
# much faster version for eos break mode
block_to_dataset_index = np.stack(
[
Mutant 2170
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -79,7 +79,7 @@
self._sizes = slice_indices[:, 1] - slice_indices[:, 0]
# build index mapping block indices to the underlying dataset indices
- if break_mode == "eos":
+ if break_mode == "XXeosXX":
# much faster version for eos break mode
block_to_dataset_index = np.stack(
[
Mutant 2181
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -124,7 +124,7 @@
)
slice_s, slice_e = self.slice_indices[index]
- length = slice_e - slice_s
+ length = slice_e + slice_s
s, e = start_offset, start_offset + length
item = buffer[s:e]
Mutant 2186
--- fairseq/data/token_block_dataset.py
+++ fairseq/data/token_block_dataset.py
@@ -151,7 +151,6 @@
def __len__(self):
return len(self.slice_indices)
- @property
def supports_prefetch(self):
return getattr(self.dataset, "supports_prefetch", False)