Skip to content

Commit 0f848ff

Browse files
author
Jesujoba Alabi
committed
updated code for quality test
1 parent 5253fb9 commit 0f848ff

31 files changed

+242
-199
lines changed

‎src/transformers/adapters/utils.py

+3-2
Original file line numberDiff line numberDiff line change
@@ -722,8 +722,9 @@ def resolve_adapter_path(
722722
except Exception as ex:
723723
logger.info(ex)
724724
raise EnvironmentError(
725-
"Unable to load adapter {} from any source. Please check the name of the adapter or the source."
726-
.format(adapter_name_or_path)
725+
"Unable to load adapter {} from any source. Please check the name of the adapter or the source.".format(
726+
adapter_name_or_path
727+
)
727728
)
728729
else:
729730
raise ValueError("Unable to identify {} as a valid module location.".format(adapter_name_or_path))

‎src/transformers/commands/add_new_model_like.py

+7-2
Original file line numberDiff line numberDiff line change
@@ -438,9 +438,14 @@ def duplicate_module(
438438
# Special cases
439439
if "PRETRAINED_CONFIG_ARCHIVE_MAP = {" in obj:
440440
# docstyle-ignore
441-
obj = f"{new_model_patterns.model_upper_cased}_PRETRAINED_CONFIG_ARCHIVE_MAP = " + "{" + f"""
441+
obj = (
442+
f"{new_model_patterns.model_upper_cased}_PRETRAINED_CONFIG_ARCHIVE_MAP = "
443+
+ "{"
444+
+ f"""
442445
"{new_model_patterns.checkpoint}": "https://huggingface.co/{new_model_patterns.checkpoint}/resolve/main/config.json",
443-
""" + "}\n"
446+
"""
447+
+ "}\n"
448+
)
444449
new_objects.append(obj)
445450
continue
446451
elif "PRETRAINED_MODEL_ARCHIVE_LIST = [" in obj:

‎src/transformers/integrations.py

+4-2
Original file line numberDiff line numberDiff line change
@@ -1046,11 +1046,13 @@ def __del__(self):
10461046

10471047
class NeptuneMissingConfiguration(Exception):
10481048
def __init__(self):
1049-
super().__init__("""
1049+
super().__init__(
1050+
"""
10501051
------ Unsupported ---- We were not able to create new runs. You provided a custom Neptune run to
10511052
`NeptuneCallback` with the `run` argument. For the integration to work fully, provide your `api_token` and
10521053
`project` by saving them as environment variables or passing them to the callback.
1053-
""")
1054+
"""
1055+
)
10541056

10551057

10561058
class NeptuneCallback(TrainerCallback):

‎src/transformers/modeling_utils.py

+4-2
Original file line numberDiff line numberDiff line change
@@ -2423,11 +2423,13 @@ def from_pretrained(cls, pretrained_model_name_or_path: Optional[Union[str, os.P
24232423
key: device_map[key] for key in device_map.keys() if key not in modules_to_not_convert
24242424
}
24252425
if "cpu" in device_map_without_lm_head.values() or "disk" in device_map_without_lm_head.values():
2426-
raise ValueError("""
2426+
raise ValueError(
2427+
"""
24272428
Some modules are dispatched on the CPU or the disk. Make sure you have enough GPU RAM to fit
24282429
the quantized model. If you have set a value for `max_memory` you should increase that. To have
24292430
an idea of the modules that are set on the CPU or RAM you can print model.hf_device_map.
2430-
""")
2431+
"""
2432+
)
24312433
del device_map_without_lm_head
24322434

24332435
if from_tf:

‎src/transformers/models/big_bird/modeling_big_bird.py

+5-3
Original file line numberDiff line numberDiff line change
@@ -927,9 +927,11 @@ def bigbird_block_sparse_attention(
927927
attention_probs[:, :, -2 * from_block_size : -from_block_size, :to_block_size] = second_last_attn_weights[
928928
:, :, :, :to_block_size
929929
] # 1st key block (global)
930-
attention_probs[:, :, -2 * from_block_size : -from_block_size, -3 * to_block_size :] = (
931-
second_last_attn_weights[:, :, :, to_block_size : 4 * to_block_size]
932-
) # last three blocks (global + sliding)
930+
attention_probs[
931+
:, :, -2 * from_block_size : -from_block_size, -3 * to_block_size :
932+
] = second_last_attn_weights[
933+
:, :, :, to_block_size : 4 * to_block_size
934+
] # last three blocks (global + sliding)
933935
# random keys
934936
for p1, i1, w1 in zip(range(bsz), rand_attn, second_last_attn_weights):
935937
# p1, i1, w1 corresponds to batch_dim i.e. following operation is done for each sequence in batch

‎src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py

+5-3
Original file line numberDiff line numberDiff line change
@@ -745,9 +745,11 @@ def bigbird_block_sparse_attention(
745745
attention_probs[:, :, -2 * from_block_size : -from_block_size, :to_block_size] = second_last_attn_weights[
746746
:, :, :, :to_block_size
747747
] # 1st key block (global)
748-
attention_probs[:, :, -2 * from_block_size : -from_block_size, -3 * to_block_size :] = (
749-
second_last_attn_weights[:, :, :, to_block_size : 4 * to_block_size]
750-
) # last three blocks (global + sliding)
748+
attention_probs[
749+
:, :, -2 * from_block_size : -from_block_size, -3 * to_block_size :
750+
] = second_last_attn_weights[
751+
:, :, :, to_block_size : 4 * to_block_size
752+
] # last three blocks (global + sliding)
751753
# random keys
752754
for p1, i1, w1 in zip(range(bsz), rand_attn, second_last_attn_weights):
753755
# p1, i1, w1 corresponds to batch_dim i.e. following operation is done for each sequence in batch

‎src/transformers/models/convbert/modeling_convbert.py

+66-66
Original file line numberDiff line numberDiff line change
@@ -88,72 +88,72 @@ def load_tf_weights_in_convbert(model, config, tf_checkpoint_path):
8888
group_dense_name = "dense"
8989

9090
for j in range(config.num_hidden_layers):
91-
param_mapping[f"encoder.layer.{j}.attention.self.query.weight"] = (
92-
f"electra/encoder/layer_{j}/attention/self/query/kernel"
93-
)
94-
param_mapping[f"encoder.layer.{j}.attention.self.query.bias"] = (
95-
f"electra/encoder/layer_{j}/attention/self/query/bias"
96-
)
97-
param_mapping[f"encoder.layer.{j}.attention.self.key.weight"] = (
98-
f"electra/encoder/layer_{j}/attention/self/key/kernel"
99-
)
100-
param_mapping[f"encoder.layer.{j}.attention.self.key.bias"] = (
101-
f"electra/encoder/layer_{j}/attention/self/key/bias"
102-
)
103-
param_mapping[f"encoder.layer.{j}.attention.self.value.weight"] = (
104-
f"electra/encoder/layer_{j}/attention/self/value/kernel"
105-
)
106-
param_mapping[f"encoder.layer.{j}.attention.self.value.bias"] = (
107-
f"electra/encoder/layer_{j}/attention/self/value/bias"
108-
)
109-
param_mapping[f"encoder.layer.{j}.attention.self.key_conv_attn_layer.depthwise.weight"] = (
110-
f"electra/encoder/layer_{j}/attention/self/conv_attn_key/depthwise_kernel"
111-
)
112-
param_mapping[f"encoder.layer.{j}.attention.self.key_conv_attn_layer.pointwise.weight"] = (
113-
f"electra/encoder/layer_{j}/attention/self/conv_attn_key/pointwise_kernel"
114-
)
115-
param_mapping[f"encoder.layer.{j}.attention.self.key_conv_attn_layer.bias"] = (
116-
f"electra/encoder/layer_{j}/attention/self/conv_attn_key/bias"
117-
)
118-
param_mapping[f"encoder.layer.{j}.attention.self.conv_kernel_layer.weight"] = (
119-
f"electra/encoder/layer_{j}/attention/self/conv_attn_kernel/kernel"
120-
)
121-
param_mapping[f"encoder.layer.{j}.attention.self.conv_kernel_layer.bias"] = (
122-
f"electra/encoder/layer_{j}/attention/self/conv_attn_kernel/bias"
123-
)
124-
param_mapping[f"encoder.layer.{j}.attention.self.conv_out_layer.weight"] = (
125-
f"electra/encoder/layer_{j}/attention/self/conv_attn_point/kernel"
126-
)
127-
param_mapping[f"encoder.layer.{j}.attention.self.conv_out_layer.bias"] = (
128-
f"electra/encoder/layer_{j}/attention/self/conv_attn_point/bias"
129-
)
130-
param_mapping[f"encoder.layer.{j}.attention.output.dense.weight"] = (
131-
f"electra/encoder/layer_{j}/attention/output/dense/kernel"
132-
)
133-
param_mapping[f"encoder.layer.{j}.attention.output.LayerNorm.weight"] = (
134-
f"electra/encoder/layer_{j}/attention/output/LayerNorm/gamma"
135-
)
136-
param_mapping[f"encoder.layer.{j}.attention.output.dense.bias"] = (
137-
f"electra/encoder/layer_{j}/attention/output/dense/bias"
138-
)
139-
param_mapping[f"encoder.layer.{j}.attention.output.LayerNorm.bias"] = (
140-
f"electra/encoder/layer_{j}/attention/output/LayerNorm/beta"
141-
)
142-
param_mapping[f"encoder.layer.{j}.intermediate.dense.weight"] = (
143-
f"electra/encoder/layer_{j}/intermediate/{group_dense_name}/kernel"
144-
)
145-
param_mapping[f"encoder.layer.{j}.intermediate.dense.bias"] = (
146-
f"electra/encoder/layer_{j}/intermediate/{group_dense_name}/bias"
147-
)
148-
param_mapping[f"encoder.layer.{j}.output.dense.weight"] = (
149-
f"electra/encoder/layer_{j}/output/{group_dense_name}/kernel"
150-
)
151-
param_mapping[f"encoder.layer.{j}.output.dense.bias"] = (
152-
f"electra/encoder/layer_{j}/output/{group_dense_name}/bias"
153-
)
154-
param_mapping[f"encoder.layer.{j}.output.LayerNorm.weight"] = (
155-
f"electra/encoder/layer_{j}/output/LayerNorm/gamma"
156-
)
91+
param_mapping[
92+
f"encoder.layer.{j}.attention.self.query.weight"
93+
] = f"electra/encoder/layer_{j}/attention/self/query/kernel"
94+
param_mapping[
95+
f"encoder.layer.{j}.attention.self.query.bias"
96+
] = f"electra/encoder/layer_{j}/attention/self/query/bias"
97+
param_mapping[
98+
f"encoder.layer.{j}.attention.self.key.weight"
99+
] = f"electra/encoder/layer_{j}/attention/self/key/kernel"
100+
param_mapping[
101+
f"encoder.layer.{j}.attention.self.key.bias"
102+
] = f"electra/encoder/layer_{j}/attention/self/key/bias"
103+
param_mapping[
104+
f"encoder.layer.{j}.attention.self.value.weight"
105+
] = f"electra/encoder/layer_{j}/attention/self/value/kernel"
106+
param_mapping[
107+
f"encoder.layer.{j}.attention.self.value.bias"
108+
] = f"electra/encoder/layer_{j}/attention/self/value/bias"
109+
param_mapping[
110+
f"encoder.layer.{j}.attention.self.key_conv_attn_layer.depthwise.weight"
111+
] = f"electra/encoder/layer_{j}/attention/self/conv_attn_key/depthwise_kernel"
112+
param_mapping[
113+
f"encoder.layer.{j}.attention.self.key_conv_attn_layer.pointwise.weight"
114+
] = f"electra/encoder/layer_{j}/attention/self/conv_attn_key/pointwise_kernel"
115+
param_mapping[
116+
f"encoder.layer.{j}.attention.self.key_conv_attn_layer.bias"
117+
] = f"electra/encoder/layer_{j}/attention/self/conv_attn_key/bias"
118+
param_mapping[
119+
f"encoder.layer.{j}.attention.self.conv_kernel_layer.weight"
120+
] = f"electra/encoder/layer_{j}/attention/self/conv_attn_kernel/kernel"
121+
param_mapping[
122+
f"encoder.layer.{j}.attention.self.conv_kernel_layer.bias"
123+
] = f"electra/encoder/layer_{j}/attention/self/conv_attn_kernel/bias"
124+
param_mapping[
125+
f"encoder.layer.{j}.attention.self.conv_out_layer.weight"
126+
] = f"electra/encoder/layer_{j}/attention/self/conv_attn_point/kernel"
127+
param_mapping[
128+
f"encoder.layer.{j}.attention.self.conv_out_layer.bias"
129+
] = f"electra/encoder/layer_{j}/attention/self/conv_attn_point/bias"
130+
param_mapping[
131+
f"encoder.layer.{j}.attention.output.dense.weight"
132+
] = f"electra/encoder/layer_{j}/attention/output/dense/kernel"
133+
param_mapping[
134+
f"encoder.layer.{j}.attention.output.LayerNorm.weight"
135+
] = f"electra/encoder/layer_{j}/attention/output/LayerNorm/gamma"
136+
param_mapping[
137+
f"encoder.layer.{j}.attention.output.dense.bias"
138+
] = f"electra/encoder/layer_{j}/attention/output/dense/bias"
139+
param_mapping[
140+
f"encoder.layer.{j}.attention.output.LayerNorm.bias"
141+
] = f"electra/encoder/layer_{j}/attention/output/LayerNorm/beta"
142+
param_mapping[
143+
f"encoder.layer.{j}.intermediate.dense.weight"
144+
] = f"electra/encoder/layer_{j}/intermediate/{group_dense_name}/kernel"
145+
param_mapping[
146+
f"encoder.layer.{j}.intermediate.dense.bias"
147+
] = f"electra/encoder/layer_{j}/intermediate/{group_dense_name}/bias"
148+
param_mapping[
149+
f"encoder.layer.{j}.output.dense.weight"
150+
] = f"electra/encoder/layer_{j}/output/{group_dense_name}/kernel"
151+
param_mapping[
152+
f"encoder.layer.{j}.output.dense.bias"
153+
] = f"electra/encoder/layer_{j}/output/{group_dense_name}/bias"
154+
param_mapping[
155+
f"encoder.layer.{j}.output.LayerNorm.weight"
156+
] = f"electra/encoder/layer_{j}/output/LayerNorm/gamma"
157157
param_mapping[f"encoder.layer.{j}.output.LayerNorm.bias"] = f"electra/encoder/layer_{j}/output/LayerNorm/beta"
158158

159159
for param in model.named_parameters():

‎src/transformers/models/donut/convert_donut_to_pytorch.py

+12-12
Original file line numberDiff line numberDiff line change
@@ -106,22 +106,22 @@ def convert_state_dict(orig_state_dict, model):
106106
orig_state_dict[
107107
f"encoder.encoder.layers.{layer_num}.blocks.{block_num}.attention.self.query.weight"
108108
] = val[:dim, :]
109-
orig_state_dict[f"encoder.encoder.layers.{layer_num}.blocks.{block_num}.attention.self.key.weight"] = (
110-
val[dim : dim * 2, :]
111-
)
109+
orig_state_dict[
110+
f"encoder.encoder.layers.{layer_num}.blocks.{block_num}.attention.self.key.weight"
111+
] = val[dim : dim * 2, :]
112112
orig_state_dict[
113113
f"encoder.encoder.layers.{layer_num}.blocks.{block_num}.attention.self.value.weight"
114114
] = val[-dim:, :]
115115
else:
116-
orig_state_dict[f"encoder.encoder.layers.{layer_num}.blocks.{block_num}.attention.self.query.bias"] = (
117-
val[:dim]
118-
)
119-
orig_state_dict[f"encoder.encoder.layers.{layer_num}.blocks.{block_num}.attention.self.key.bias"] = (
120-
val[dim : dim * 2]
121-
)
122-
orig_state_dict[f"encoder.encoder.layers.{layer_num}.blocks.{block_num}.attention.self.value.bias"] = (
123-
val[-dim:]
124-
)
116+
orig_state_dict[
117+
f"encoder.encoder.layers.{layer_num}.blocks.{block_num}.attention.self.query.bias"
118+
] = val[:dim]
119+
orig_state_dict[
120+
f"encoder.encoder.layers.{layer_num}.blocks.{block_num}.attention.self.key.bias"
121+
] = val[dim : dim * 2]
122+
orig_state_dict[
123+
f"encoder.encoder.layers.{layer_num}.blocks.{block_num}.attention.self.value.bias"
124+
] = val[-dim:]
125125
elif "attn_mask" in key or key in ["encoder.model.norm.weight", "encoder.model.norm.bias"]:
126126
# HuggingFace implementation doesn't use attn_mask buffer
127127
# and model doesn't use final LayerNorms for the encoder

‎src/transformers/models/flava/modeling_flava.py

+17-6
Original file line numberDiff line numberDiff line change
@@ -775,11 +775,14 @@ def forward(self, hidden_states: torch.Tensor):
775775

776776
FLAVA_TEXT_INPUTS_DOCSTRING = FLAVA_TEXT_INPUTS_DOCSTRING_BASE + FLAVA_INPUTS_DOCSTRING_COMMON
777777

778-
FLAVA_MULTIMODAL_INPUTS_DOCSTRING = r"""
778+
FLAVA_MULTIMODAL_INPUTS_DOCSTRING = (
779+
r"""
779780
Args:
780781
hidden_states (`torch.FloatTensor` of shape `(batch_size, image_num_patches + text_seq_len, hidden_size)`):
781782
The concatenated hidden states of unimodal encoders.
782-
""" + FLAVA_INPUTS_DOCSTRING_COMMON
783+
"""
784+
+ FLAVA_INPUTS_DOCSTRING_COMMON
785+
)
783786

784787
FLAVA_MODEL_INPUTS_DOCSTRING_BASE = r"""
785788
Args:
@@ -1260,7 +1263,9 @@ def get_text_features(
12601263
... text=["a photo of a cat", "a photo of a dog"], max_length=77, padding="max_length", return_tensors="pt"
12611264
... )
12621265
>>> text_features = model.get_text_features(**inputs)
1263-
```""".format(_CHECKPOINT_FOR_DOC)
1266+
```""".format(
1267+
_CHECKPOINT_FOR_DOC
1268+
)
12641269
text_outputs = self.text_model(
12651270
input_ids=input_ids,
12661271
attention_mask=attention_mask,
@@ -1309,7 +1314,9 @@ def get_image_features(
13091314
>>> inputs = processor(images=image, return_tensors="pt")
13101315
13111316
>>> image_features = model.get_image_features(**inputs)
1312-
```""".format(_CHECKPOINT_FOR_DOC)
1317+
```""".format(
1318+
_CHECKPOINT_FOR_DOC
1319+
)
13131320
image_outputs = self.image_model(
13141321
pixel_values=pixel_values,
13151322
bool_masked_pos=bool_masked_pos,
@@ -1574,7 +1581,9 @@ def get_codebook_indices(self, pixel_values: torch.Tensor) -> torch.Tensor:
15741581
15751582
>>> outputs = model.get_codebook_indices(**inputs)
15761583
```
1577-
""".format(_CHECKPOINT_FOR_CODEBOOK_DOC)
1584+
""".format(
1585+
_CHECKPOINT_FOR_CODEBOOK_DOC
1586+
)
15781587
z_logits = self.blocks(pixel_values)
15791588
return torch.argmax(z_logits, axis=1)
15801589

@@ -1609,7 +1618,9 @@ def forward(self, pixel_values: torch.FloatTensor) -> torch.Tensor:
16091618
>>> print(outputs.shape)
16101619
(1, 196)
16111620
```
1612-
""".format(_CHECKPOINT_FOR_CODEBOOK_DOC)
1621+
""".format(
1622+
_CHECKPOINT_FOR_CODEBOOK_DOC
1623+
)
16131624
if len(pixel_values.shape) != 4:
16141625
raise ValueError(f"input shape {pixel_values.shape} is not 4d")
16151626
if pixel_values.shape[1] != self.input_channels:

‎src/transformers/models/layoutlmv2/tokenization_layoutlmv2.py

+1-2
Original file line numberDiff line numberDiff line change
@@ -1213,8 +1213,7 @@ def truncate_sequences(
12131213
)
12141214
if truncation_strategy == TruncationStrategy.ONLY_FIRST:
12151215
error_msg = (
1216-
error_msg
1217-
+ "Please select another truncation strategy than "
1216+
error_msg + "Please select another truncation strategy than "
12181217
f"{truncation_strategy}, for instance 'longest_first' or 'only_second'."
12191218
)
12201219
logger.error(error_msg)

‎src/transformers/models/layoutlmv3/tokenization_layoutlmv3.py

+1-2
Original file line numberDiff line numberDiff line change
@@ -1345,8 +1345,7 @@ def truncate_sequences(
13451345
)
13461346
if truncation_strategy == TruncationStrategy.ONLY_FIRST:
13471347
error_msg = (
1348-
error_msg
1349-
+ "Please select another truncation strategy than "
1348+
error_msg + "Please select another truncation strategy than "
13501349
f"{truncation_strategy}, for instance 'longest_first' or 'only_second'."
13511350
)
13521351
logger.error(error_msg)

‎src/transformers/models/marian/convert_marian_tatoeba_to_pytorch.py

+12-2
Original file line numberDiff line numberDiff line change
@@ -236,14 +236,24 @@ def write_model_card(self, model_dict, dry_run=False) -> str:
236236
* OPUS readme: [README.md]({readme_url})
237237
"""
238238

239-
content = f"""
239+
content = (
240+
f"""
240241
* model: {model_dict['modeltype']}
241242
* source language code{src_multilingual*'s'}: {', '.join(a2_src_tags)}
242243
* target language code{tgt_multilingual*'s'}: {', '.join(a2_tgt_tags)}
243244
* dataset: opus {backtranslated_data}
244245
* release date: {model_dict['release-date']}
245246
* pre-processing: {model_dict['pre-processing']}
246-
""" + multilingual_data + tuned + download + langtoken + datainfo + testset + testscores + scorestable
247+
"""
248+
+ multilingual_data
249+
+ tuned
250+
+ download
251+
+ langtoken
252+
+ datainfo
253+
+ testset
254+
+ testscores
255+
+ scorestable
256+
)
247257

248258
content = FRONT_MATTER_TEMPLATE.format(lang_tags) + extra_markdown + content
249259

‎src/transformers/models/markuplm/tokenization_markuplm.py

+1-2
Original file line numberDiff line numberDiff line change
@@ -1315,8 +1315,7 @@ def truncate_sequences(
13151315
)
13161316
if truncation_strategy == TruncationStrategy.ONLY_FIRST:
13171317
error_msg = (
1318-
error_msg
1319-
+ "Please select another truncation strategy than "
1318+
error_msg + "Please select another truncation strategy than "
13201319
f"{truncation_strategy}, for instance 'longest_first' or 'only_second'."
13211320
)
13221321
logger.error(error_msg)

0 commit comments

Comments
 (0)