def test_beam_search(self):
vocab = text_utils.VocabFromText(self.VOCAB_EXAMPLE_SENTENCES)
model_config = self.config.model_config.butd
model = TestDecoderModel(model_config, vocab)
model.build()
model.eval()
expected_tokens = {
1: [1.0, 23.0, 1.0, 24.0, 29.0, 37.0, 40.0, 17.0, 29.0, 2.0],
2: [1.0, 0.0, 8.0, 1.0, 28.0, 25.0, 2.0],
8: [1.0, 34.0, 1.0, 13.0, 1.0, 2.0],
16: [1.0, 25.0, 18.0, 2.0],
}
for batch_size in [1, 2, 8, 16]:
samples = []
for _ in range(batch_size):
sample = Sample()
sample.dataset_name = "coco"
sample.dataset_type = "test"
sample.image_feature_0 = torch.randn(100, 2048)
sample.answers = torch.zeros((5, 10), dtype=torch.long)
samples.append(sample)
sample_list = SampleList(samples)
tokens = model(sample_list)["captions"]
self.assertEqual(np.trim_zeros(tokens[0].tolist()),
expected_tokens[batch_size])
def compare_torchscript_transformer_models(model, vocab_size):
test_sample = Sample()
test_sample.input_ids = torch.randint(low=0, high=vocab_size,
size=(128, )).long()
test_sample.input_mask = torch.ones(128).long()
test_sample.segment_ids = torch.zeros(128).long()
test_sample.image_feature_0 = torch.rand((1, 100, 2048)).float()
test_sample.image = torch.rand((3, 300, 300)).float()
test_sample_list = SampleList([test_sample])
with torch.no_grad():
model_output = model(test_sample_list)
script_model = torch.jit.script(model)
with torch.no_grad():
script_output = script_model(test_sample_list)
return torch.equal(model_output["scores"], script_output["scores"])
def test_nucleus_sampling(self):
vocab = text_utils.VocabFromText(self.VOCAB_EXAMPLE_SENTENCES)
model_config = self.config.model_config.butd
model = TestDecoderModel(model_config, vocab)
model.build()
model.eval()
sample = Sample()
sample.dataset_name = "coco"
sample.dataset_type = "test"
sample.image_feature_0 = torch.randn(100, 2048)
sample.answers = torch.zeros((5, 10), dtype=torch.long)
sample_list = SampleList([sample])
tokens = model(sample_list)["captions"]
# these are expected tokens for sum_threshold = 0.5
# Because of a bug fix in https://github.com/pytorch/pytorch/pull/47386
# the torch.Tensor.multinomail will generate different random sequence.
# TODO: Remove this hack after OSS uses later version of PyTorch.
if LegacyVersion(torch.__version__) > LegacyVersion("1.7.1"):
expected_tokens = [1.0, 23.0, 38.0, 30.0, 5.0, 11.0, 2.0]
else:
expected_tokens = [
1.0,
29.0,
11.0,
11.0,
39.0,
10.0,
31.0,
4.0,
19.0,
39.0,
2.0,
]
self.assertEqual(tokens[0].tolist(), expected_tokens)