Skip to content

cleanup huggingface tests #1370

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Feb 26, 2021
Merged

cleanup huggingface tests #1370

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b7c1e8c
cleanup huggingface tests
guschmue Feb 26, 2021
199e88d
Merge branch 'master' into gs/cleanup-hugg
TomWildenhain-Microsoft Feb 26, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
111 changes: 63 additions & 48 deletions tests/huggingface.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -69,7 +69,8 @@ def run_keras(self, model, inputs):
print(f'==== avg keras name={self.name}, time={val}')
return pred

def run_test(self, model, input_dict, rtol=1e-2, atol=1e-4, input_signature=None, outputs=None, large=True):
def run_test(self, model, input_dict, rtol=1e-2, atol=1e-4, input_signature=None,
outputs=None, large=True, extra_input=None):

# always use external model format for consistency
large = True
Expand All @@ -94,6 +95,7 @@ def run_test(self, model, input_dict, rtol=1e-2, atol=1e-4, input_signature=None
model_path = os.path.join(dst, self.name)
if not large:
model_path = model_path + ".onnx"
print("= convert")
_, _ = tf2onnx.convert.from_keras(model, input_signature=input_signature,
opset=13, large_model=large, output_path=model_path)

Expand All @@ -103,9 +105,28 @@ def run_test(self, model, input_dict, rtol=1e-2, atol=1e-4, input_signature=None
z.extractall(os.path.dirname(model_path))
model_path = os.path.join(os.path.dirname(model_path), "__MODEL_PROTO.onnx")

print("= running ort")
if extra_input:
input_dict.update(extra_input)
onnx_results = self.run_onnxruntime(model_path, input_dict, outputs)
self.assertAllClose(tf_results, onnx_results, rtol=rtol, atol=atol)

def spec_and_pad(self, input_dict, max_length=None, batchdim=None):
spec = []
new_dict = {}
for k, v in input_dict.items():
shape = v.shape
if len(shape) == 2:
if not max_length:
shape = [batchdim, None]
else:
shape = [batchdim, max_length]
spec.append(tf.TensorSpec(shape, dtype=v.dtype, name=k))
if max_length:
l = len(v[0])
v = tf.pad(v, [[0, 0], [0, max_length-l]])
new_dict[k] = v
return tuple(spec), new_dict

# BERT

Expand All @@ -115,9 +136,7 @@ def test_TFBertModel(self):
model = TFBertForQuestionAnswering.from_pretrained('bert-base-cased')
question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet"
input_dict = tokenizer(question, text, return_tensors='tf')
spec = (tf.TensorSpec((None, None), tf.int32, name="input_ids"),
tf.TensorSpec((None, None), tf.int32, name="token_type_ids"),
tf.TensorSpec((None, None), tf.int32, name="attention_mask"))
spec, input_dict = self.spec_and_pad(input_dict)
self.run_test(model, input_dict, input_signature=spec)

def test_TFBertFineTunedSquadModel(self):
Expand All @@ -127,9 +146,7 @@ def test_TFBertFineTunedSquadModel(self):
model = TFBertForQuestionAnswering.from_pretrained(name)
question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet"
input_dict = tokenizer(question, text, return_tensors='tf')
spec = (tf.TensorSpec((None, None), tf.int32, name="input_ids"),
tf.TensorSpec((None, None), tf.int32, name="token_type_ids"),
tf.TensorSpec((None, None), tf.int32, name="attention_mask"))
spec, input_dict = self.spec_and_pad(input_dict)
self.run_test(model, input_dict, input_signature=spec)

def test_TFDisillBertModel(self):
Expand All @@ -140,8 +157,7 @@ def test_TFDisillBertModel(self):
model = TFDistilBertForQuestionAnswering.from_pretrained(name)
question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet"
input_dict = tokenizer(question, text, return_tensors='tf')
spec = (tf.TensorSpec((None, None), tf.int32, name="input_ids"),
tf.TensorSpec((None, None), tf.int32, name="attention_mask"))
spec, input_dict = self.spec_and_pad(input_dict)
outputs = ["start_logits", "end_logits"]
self.run_test(model, input_dict, input_signature=spec, outputs=outputs, rtol=1e-5)

Expand All @@ -153,9 +169,7 @@ def _test_TFFunnelSquad(self, size, large=False):
model = TFFunnelForQuestionAnswering.from_pretrained(size)
question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet"
input_dict = tokenizer(question, text, return_tensors='tf')
spec = (tf.TensorSpec((None, 14), tf.int32, name="input_ids"),
tf.TensorSpec((None, 14), tf.int32, name="token_type_ids"),
tf.TensorSpec((None, 14), tf.int32, name="attention_mask"))
spec, input_dict = self.spec_and_pad(input_dict, max_length=model.config.max_length)
outputs = ["start_logits", "end_logits"]
self.run_test(model, input_dict, input_signature=spec, outputs=outputs, rtol=1e-5)

Expand All @@ -172,10 +186,10 @@ def _test_TFT5Model(self, size, large=False):
tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="tf").input_ids
decoder_input_ids = \
tokenizer("Studies show that", return_tensors="tf").input_ids
spec = (tf.TensorSpec((None, None), tf.int32, name="input_ids"),
tf.TensorSpec((None, None), tf.int32, name="decoder_input_ids"))
input_dict = {"input_ids": input_ids, "decoder_input_ids": decoder_input_ids}
spec, input_dict = self.spec_and_pad(input_dict)
outputs = ["last_hidden_state"]
self.run_test(model, {"input_ids": input_ids, "decoder_input_ids": decoder_input_ids},
self.run_test(model, input_dict,
input_signature=spec, outputs=outputs, large=large)

def test_TFT5ModelSmall(self):
Expand All @@ -199,12 +213,10 @@ def _test_TFAlbert(self, size, large=False):
from transformers import AlbertTokenizer, TFAlbertModel
tokenizer = AlbertTokenizer.from_pretrained(size)
model = TFAlbertModel.from_pretrained(size)
inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")
spec = (tf.TensorSpec((None, None), tf.int32, name="input_ids"),
tf.TensorSpec((None, None), tf.int32, name="token_type_ids"),
tf.TensorSpec((None, None), tf.int32, name="attention_mask"))
input_dict = tokenizer("Hello, my dog is cute", return_tensors="tf")
spec, input_dict = self.spec_and_pad(input_dict)
outputs = ["last_hidden_state"]
self.run_test(model, inputs, input_signature=spec, outputs=outputs, large=large)
self.run_test(model, input_dict, input_signature=spec, outputs=outputs, large=large)

def test_TFAlbertBaseV1(self):
self._test_TFAlbert("albert-base-v1", large=True)
Expand Down Expand Up @@ -236,21 +248,19 @@ def test_TFCTRL(self):
from transformers import CTRLTokenizer, TFCTRLModel
tokenizer = CTRLTokenizer.from_pretrained('ctrl')
model = TFCTRLModel.from_pretrained('ctrl')
inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")
spec = (tf.TensorSpec((None, None), tf.int32, name="input_ids"),
tf.TensorSpec((None, None), tf.int32, name="attention_mask"))
input_dict = tokenizer("Hello, my dog is cute", return_tensors="tf")
spec, input_dict = self.spec_and_pad(input_dict)
outputs = ["last_hidden_state"]
self.run_test(model, inputs, input_signature=spec, outputs=outputs, large=True)
self.run_test(model, input_dict, input_signature=spec, outputs=outputs, large=True)

def _test_TFGpt2(self, size, large=False):
from transformers import GPT2Tokenizer, TFGPT2Model
tokenizer = GPT2Tokenizer.from_pretrained(size)
model = TFGPT2Model.from_pretrained(size)
inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")
spec = (tf.TensorSpec((None, None), tf.int32, name="input_ids"),
tf.TensorSpec((None, None), tf.int32, name="attention_mask"))
input_dict = tokenizer("Hello, my dog is cute", return_tensors="tf")
spec, input_dict = self.spec_and_pad(input_dict)
outputs = ["last_hidden_state"]
self.run_test(model, inputs, input_signature=spec, outputs=outputs, large=large)
self.run_test(model, input_dict, input_signature=spec, outputs=outputs, large=large)

# GPT2

Expand All @@ -272,13 +282,18 @@ def _test_TFLongformer(self, size, large=False):
from transformers import LongformerTokenizer, TFLongformerModel
tokenizer = LongformerTokenizer.from_pretrained(size)
model = TFLongformerModel.from_pretrained(size)
inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")
spec = (tf.TensorSpec((1, 8), tf.int32, name="input_ids"),
tf.TensorSpec((1, 8), tf.int32, name="attention_mask"))
input_dict = tokenizer("Hello, my dog is cute", return_tensors="tf")
spec, input_dict = self.spec_and_pad(input_dict, max_length=512)
outputs = ["last_hidden_state"]
self.run_test(model, inputs, input_signature=spec, outputs=outputs, large=large)
self.run_test(model, input_dict, input_signature=spec, outputs=outputs, large=large)

def test_TFLongformerBase(self):
# fails since transformers-2.4.2?
#
# transformers/models/longformer/modeling_tf_longformer.py", line 1839, in _pad_to_window_size
# if tf.math.greater(padding_len, 0)
# OperatorNotAllowedInGraphError: using a `tf.Tensor` as a Python `bool` is not allowed
#
self._test_TFLongformer("allenai/longformer-base-4096", large=True)

def test_TFLongformerLarge(self):
Expand All @@ -295,11 +310,17 @@ def _test_TFPegasus(self, size, large=False):
decoder_input_ids = \
tokenizer("Studies show that", return_tensors="tf").input_ids

spec = (tf.TensorSpec((1, 13), tf.int32, name="input_ids"),
tf.TensorSpec((1, 4), tf.int32, name="decoder_input_ids"))
input_dict = {"input_ids": input_ids, "decoder_input_ids": decoder_input_ids}

# this comes from TFPegasusEncoder/Decoder like:
# self.embed_scale = tf.math.sqrt(float(config.d_model)) if config.scale_embedding else 1.0
# while this is mean to come from config tf tells us that those are model inputs
# this might be new in tensformers-2.4.2, we did not notice before that
extra_input = {"tf_pegasus_model/model/decoder/mul/y:0": np.array([32.], dtype=np.float32),
"tf_pegasus_model/model/encoder/mul/y:0": np.array([32.], dtype=np.float32)}
spec, input_dict = self.spec_and_pad(input_dict, max_length=model.config.max_length)
outputs = ["last_hidden_state"]
self.run_test(model, {"input_ids": input_ids, "decoder_input_ids": decoder_input_ids},
input_signature=spec, outputs=outputs, large=large)
self.run_test(model, input_dict, input_signature=spec, outputs=outputs, large=large, extra_input=extra_input)

def test_TFPegasus(self):
self._test_TFPegasus("google/pegasus-xsum", large=True)
Expand All @@ -310,13 +331,10 @@ def _test_TFXLM(self, size, large=False):
from transformers import TFXLMModel, XLMTokenizer
tokenizer = XLMTokenizer.from_pretrained(size)
model = TFXLMModel.from_pretrained(size)
inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")
input_size = None # tokenizer.max_model_input_sizes[size]
spec = (tf.TensorSpec((None, input_size), tf.int32, name="input_ids"),
tf.TensorSpec((None, input_size), tf.int32, name="token_type_ids"),
tf.TensorSpec((None, input_size), tf.int32, name="attention_mask"))
input_dict = tokenizer("Hello, my dog is cute", return_tensors="tf")
spec, input_dict = self.spec_and_pad(input_dict)
outputs = ["last_hidden_state"]
self.run_test(model, inputs, input_signature=spec, outputs=outputs, large=large, atol=0.005)
self.run_test(model, input_dict, input_signature=spec, outputs=outputs, large=large, atol=0.005)

def test_TFXLM(self):
self._test_TFXLM("xlm-mlm-en-2048", large=True)
Expand All @@ -333,13 +351,10 @@ def _test_TFBart(self, size, large=False):
from transformers import BartTokenizer, TFBartModel
tokenizer = BartTokenizer.from_pretrained(size)
model = TFBartModel.from_pretrained(size)
inputs = tokenizer("Hello, my dog is cute", return_tensors="tf")
print(inputs)
l = 8
spec = (tf.TensorSpec((1, l), tf.int32, name="input_ids"),
tf.TensorSpec((1, l), tf.int32, name="attention_mask"))
input_dict = tokenizer("Hello, my dog is cute", return_tensors="tf")
spec, input_dict = self.spec_and_pad(input_dict, max_length=model.config.max_length)
outputs = ["last_hidden_state"]
self.run_test(model, inputs, input_signature=spec, outputs=outputs, large=large)
self.run_test(model, input_dict, input_signature=spec, outputs=outputs, large=large)

def test_TFBartBase(self):
self._test_TFBart("facebook/bart-base", large=True)
Expand Down