From c126a239bcea9c68453cf86045a5177afbe2be6c Mon Sep 17 00:00:00 2001 From: Matt Date: Mon, 12 Sep 2022 17:51:10 +0100 Subject: [PATCH] Fix tflongformer int dtype (#18907) * Use int64 throughout TFLongFormer * make style * Do some more fixed casting in TFLongFormer * Fix some wonky "is None" conditionals * Cast all the dtypes, salt the earth * Fix copies to TFLED as well and do some casting there * dtype fix in TFLongformer test * Make fixup * Expand tolerances on the LED tests too (I think this is a TF32 thing) * Expand test tolerances for LED a tiny bit (probably a Tensorfloat thing again) --- .../models/led/modeling_tf_led.py | 16 +-- .../longformer/modeling_tf_longformer.py | 95 ++++++++++++---- tests/models/led/test_modeling_tf_led.py | 4 +- .../longformer/test_modeling_tf_longformer.py | 106 +++++++++--------- 4 files changed, 137 insertions(+), 84 deletions(-) diff --git a/src/transformers/models/led/modeling_tf_led.py b/src/transformers/models/led/modeling_tf_led.py index 6435516fb5..c677581635 100644 --- a/src/transformers/models/led/modeling_tf_led.py +++ b/src/transformers/models/led/modeling_tf_led.py @@ -472,7 +472,7 @@ class TFLEDEncoderSelfAttention(tf.keras.layers.Layer): ) first_chunk_mask = ( tf.tile( - tf.range(chunks_count + 1)[None, :, None, None], + tf.range(chunks_count + 1, dtype=tf.int64)[None, :, None, None], (batch_size * num_heads, 1, window_overlap, window_overlap), ) < 1 @@ -1335,10 +1335,10 @@ class TFLEDPreTrainedModel(TFPreTrainedModel): @property def dummy_inputs(self): - input_ids = tf.convert_to_tensor([[7, 6, 0, 0, 1], [1, 2, 3, 0, 0]]) + input_ids = tf.convert_to_tensor([[7, 6, 0, 0, 1], [1, 2, 3, 0, 0]], dtype=tf.int64) # make sure global layers are initialized - attention_mask = tf.convert_to_tensor([[1, 1, 0, 0, 1], [1, 1, 1, 0, 0]]) - global_attention_mask = tf.convert_to_tensor([[0, 0, 0, 0, 1], [0, 0, 1, 0, 0]]) + attention_mask = tf.convert_to_tensor([[1, 1, 0, 0, 1], [1, 1, 1, 0, 0]], dtype=tf.int64) + global_attention_mask = tf.convert_to_tensor([[0, 0, 0, 0, 1], [0, 0, 1, 0, 0]], dtype=tf.int64) dummy_inputs = { "input_ids": input_ids, "attention_mask": attention_mask, @@ -1350,10 +1350,10 @@ class TFLEDPreTrainedModel(TFPreTrainedModel): @tf.function( input_signature=[ { - "input_ids": tf.TensorSpec((None, None), tf.int32, name="input_ids"), - "attention_mask": tf.TensorSpec((None, None), tf.int32, name="attention_mask"), - "decoder_input_ids": tf.TensorSpec((None, None), tf.int32, name="decoder_input_ids"), - "decoder_attention_mask": tf.TensorSpec((None, None), tf.int32, name="decoder_attention_mask"), + "input_ids": tf.TensorSpec((None, None), tf.int64, name="input_ids"), + "attention_mask": tf.TensorSpec((None, None), tf.int64, name="attention_mask"), + "decoder_input_ids": tf.TensorSpec((None, None), tf.int64, name="decoder_input_ids"), + "decoder_attention_mask": tf.TensorSpec((None, None), tf.int64, name="decoder_attention_mask"), } ] ) diff --git a/src/transformers/models/longformer/modeling_tf_longformer.py b/src/transformers/models/longformer/modeling_tf_longformer.py index eab0d80054..6b491638cc 100644 --- a/src/transformers/models/longformer/modeling_tf_longformer.py +++ b/src/transformers/models/longformer/modeling_tf_longformer.py @@ -395,11 +395,10 @@ def _compute_global_attention_mask(input_ids_shape, sep_token_indices, before_se Computes global attention mask by putting attention on all tokens before `sep_token_id` if `before_sep_token is True` else after `sep_token_id`. """ - assert shape_list(sep_token_indices)[1] == 2, "`input_ids` should have two dimensions" question_end_index = tf.reshape(sep_token_indices, (input_ids_shape[0], 3, 2))[:, 0, 1][:, None] # bool attention mask with True in locations of global attention - attention_mask = tf.expand_dims(tf.range(input_ids_shape[1]), axis=0) + attention_mask = tf.expand_dims(tf.range(input_ids_shape[1], dtype=tf.int64), axis=0) attention_mask = tf.tile(attention_mask, (input_ids_shape[0], 1)) if before_sep_token is True: question_end_index = tf.tile(question_end_index, (1, input_ids_shape[1])) @@ -468,10 +467,9 @@ class TFLongformerLMHead(tf.keras.layers.Layer): return hidden_states -# Copied from transformers.models.roberta.modeling_tf_roberta.TFRobertaEmbeddings with Roberta->Longformer class TFLongformerEmbeddings(tf.keras.layers.Layer): """ - Same as BertEmbeddings with a tiny tweak for positional embeddings indexing. + Same as BertEmbeddings with a tiny tweak for positional embeddings indexing and some extra casting. """ def __init__(self, config, **kwargs): @@ -547,7 +545,7 @@ class TFLongformerEmbeddings(tf.keras.layers.Layer): input_shape = shape_list(inputs_embeds)[:-1] if token_type_ids is None: - token_type_ids = tf.fill(dims=input_shape, value=0) + token_type_ids = tf.cast(tf.fill(dims=input_shape, value=0), tf.int64) if position_ids is None: if input_ids is not None: @@ -557,7 +555,8 @@ class TFLongformerEmbeddings(tf.keras.layers.Layer): ) else: position_ids = tf.expand_dims( - tf.range(start=self.padding_idx + 1, limit=input_shape[-1] + self.padding_idx + 1), axis=0 + tf.range(start=self.padding_idx + 1, limit=input_shape[-1] + self.padding_idx + 1, dtype=tf.int64), + axis=0, ) position_embeds = tf.gather(params=self.position_embeddings, indices=position_ids) @@ -998,7 +997,7 @@ class TFLongformerSelfAttention(tf.keras.layers.Layer): ) first_chunk_mask = ( tf.tile( - tf.range(chunks_count + 1)[None, :, None, None], + tf.range(chunks_count + 1, dtype=tf.int64)[None, :, None, None], (batch_size * num_heads, 1, window_overlap, window_overlap), ) < 1 @@ -1701,6 +1700,21 @@ class TFLongformerMainLayer(tf.keras.layers.Layer): training=False, ): + if input_ids is not None and not isinstance(input_ids, tf.Tensor): + input_ids = tf.convert_to_tensor(input_ids, dtype=tf.int64) + elif input_ids is not None: + input_ids = tf.cast(input_ids, tf.int64) + + if attention_mask is not None and not isinstance(attention_mask, tf.Tensor): + attention_mask = tf.convert_to_tensor(attention_mask, dtype=tf.int64) + elif attention_mask is not None: + attention_mask = tf.cast(attention_mask, tf.int64) + + if global_attention_mask is not None and not isinstance(global_attention_mask, tf.Tensor): + global_attention_mask = tf.convert_to_tensor(global_attention_mask, dtype=tf.int64) + elif global_attention_mask is not None: + global_attention_mask = tf.cast(global_attention_mask, tf.int64) + if input_ids is not None and inputs_embeds is not None: raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time") elif input_ids is not None: @@ -1711,10 +1725,10 @@ class TFLongformerMainLayer(tf.keras.layers.Layer): raise ValueError("You have to specify either input_ids or inputs_embeds") if attention_mask is None: - attention_mask = tf.fill(input_shape, 1) + attention_mask = tf.cast(tf.fill(input_shape, 1), tf.int64) if token_type_ids is None: - token_type_ids = tf.fill(input_shape, 0) + token_type_ids = tf.cast(tf.fill(input_shape, 0), tf.int64) # merge `global_attention_mask` and `attention_mask` if global_attention_mask is not None: @@ -1831,7 +1845,7 @@ class TFLongformerMainLayer(tf.keras.layers.Layer): if inputs_embeds is not None: def pad_embeddings(): - input_ids_padding = tf.fill((batch_size, padding_len), self.pad_token_id) + input_ids_padding = tf.cast(tf.fill((batch_size, padding_len), self.pad_token_id), tf.int64) inputs_embeds_padding = self.embeddings(input_ids_padding) return tf.concat([inputs_embeds, inputs_embeds_padding], axis=-2) @@ -1875,10 +1889,15 @@ class TFLongformerPreTrainedModel(TFPreTrainedModel): @property def dummy_inputs(self): - input_ids = tf.convert_to_tensor([[7, 6, 0, 0, 1], [1, 2, 3, 0, 0], [0, 0, 0, 4, 5]]) + input_ids = tf.convert_to_tensor([[7, 6, 0, 0, 1], [1, 2, 3, 0, 0], [0, 0, 0, 4, 5]], dtype=tf.int64) # make sure global layers are initialized - attention_mask = tf.convert_to_tensor([[1, 1, 0, 0, 1], [1, 1, 1, 0, 0], [1, 0, 0, 1, 1]]) - global_attention_mask = tf.convert_to_tensor([[0, 0, 0, 0, 1], [0, 0, 1, 0, 0], [0, 0, 0, 0, 1]]) + attention_mask = tf.convert_to_tensor([[1, 1, 0, 0, 1], [1, 1, 1, 0, 0], [1, 0, 0, 1, 1]], dtype=tf.int64) + global_attention_mask = tf.convert_to_tensor( + [[0, 0, 0, 0, 1], [0, 0, 1, 0, 0], [0, 0, 0, 0, 1]], dtype=tf.int64 + ) + global_attention_mask = tf.convert_to_tensor( + [[0, 0, 0, 0, 1], [0, 0, 1, 0, 0], [0, 0, 0, 0, 1]], dtype=tf.int64 + ) return { "input_ids": input_ids, "attention_mask": attention_mask, @@ -1888,8 +1907,8 @@ class TFLongformerPreTrainedModel(TFPreTrainedModel): @tf.function( input_signature=[ { - "input_ids": tf.TensorSpec((None, None), tf.int32, name="input_ids"), - "attention_mask": tf.TensorSpec((None, None), tf.int32, name="attention_mask"), + "input_ids": tf.TensorSpec((None, None), tf.int64, name="input_ids"), + "attention_mask": tf.TensorSpec((None, None), tf.int64, name="attention_mask"), } ] ) @@ -2235,6 +2254,21 @@ class TFLongformerForQuestionAnswering(TFLongformerPreTrainedModel, TFQuestionAn are not taken into account for computing the loss. """ + if input_ids is not None and not isinstance(input_ids, tf.Tensor): + input_ids = tf.convert_to_tensor(input_ids, dtype=tf.int64) + elif input_ids is not None: + input_ids = tf.cast(input_ids, tf.int64) + + if attention_mask is not None and not isinstance(attention_mask, tf.Tensor): + attention_mask = tf.convert_to_tensor(attention_mask, dtype=tf.int64) + elif attention_mask is not None: + attention_mask = tf.cast(attention_mask, tf.int64) + + if global_attention_mask is not None and not isinstance(global_attention_mask, tf.Tensor): + global_attention_mask = tf.convert_to_tensor(global_attention_mask, dtype=tf.int64) + elif global_attention_mask is not None: + global_attention_mask = tf.cast(global_attention_mask, tf.int64) + # set global attention on question tokens if global_attention_mask is None and input_ids is not None: if shape_list(tf.where(input_ids == self.config.sep_token_id))[0] != 3 * shape_list(input_ids)[0]: @@ -2244,12 +2278,12 @@ class TFLongformerForQuestionAnswering(TFLongformerPreTrainedModel, TFQuestionAn " forward function to avoid this. This is most likely an error. The global attention is disabled" " for this forward pass." ) - global_attention_mask = tf.fill(shape_list(input_ids), value=0) + global_attention_mask = tf.cast(tf.fill(shape_list(input_ids), value=0), tf.int64) else: logger.info("Initializing global attention on question tokens...") # put global attention on all tokens until `config.sep_token_id` is reached sep_token_indices = tf.where(input_ids == self.config.sep_token_id) - sep_token_indices = tf.cast(sep_token_indices, dtype=input_ids.dtype) + sep_token_indices = tf.cast(sep_token_indices, dtype=tf.int64) global_attention_mask = _compute_global_attention_mask(shape_list(input_ids), sep_token_indices) outputs = self.longformer( @@ -2375,13 +2409,28 @@ class TFLongformerForSequenceClassification(TFLongformerPreTrainedModel, TFSeque training: Optional[bool] = False, ) -> Union[TFLongformerSequenceClassifierOutput, Tuple[tf.Tensor]]: + if input_ids is not None and not isinstance(input_ids, tf.Tensor): + input_ids = tf.convert_to_tensor(input_ids, dtype=tf.int64) + elif input_ids is not None: + input_ids = tf.cast(input_ids, tf.int64) + + if attention_mask is not None and not isinstance(attention_mask, tf.Tensor): + attention_mask = tf.convert_to_tensor(attention_mask, dtype=tf.int64) + elif attention_mask is not None: + attention_mask = tf.cast(attention_mask, tf.int64) + + if global_attention_mask is not None and not isinstance(global_attention_mask, tf.Tensor): + global_attention_mask = tf.convert_to_tensor(global_attention_mask, dtype=tf.int64) + elif global_attention_mask is not None: + global_attention_mask = tf.cast(global_attention_mask, tf.int64) + if global_attention_mask is None and input_ids is not None: logger.info("Initializing global attention on CLS token...") # global attention on cls token global_attention_mask = tf.zeros_like(input_ids) - updates = tf.ones(shape_list(input_ids)[0], dtype=tf.int32) + updates = tf.ones(shape_list(input_ids)[0], dtype=tf.int64) indices = tf.pad( - tensor=tf.expand_dims(tf.range(shape_list(input_ids)[0]), axis=1), + tensor=tf.expand_dims(tf.range(shape_list(input_ids)[0], dtype=tf.int64), axis=1), paddings=[[0, 0], [0, 1]], constant_values=0, ) @@ -2453,9 +2502,9 @@ class TFLongformerForMultipleChoice(TFLongformerPreTrainedModel, TFMultipleChoic @property def dummy_inputs(self): - input_ids = tf.convert_to_tensor(MULTIPLE_CHOICE_DUMMY_INPUTS) + input_ids = tf.convert_to_tensor(MULTIPLE_CHOICE_DUMMY_INPUTS, dtype=tf.int64) # make sure global layers are initialized - global_attention_mask = tf.convert_to_tensor([[[0, 0, 0, 1], [0, 0, 0, 1]]] * 2) + global_attention_mask = tf.convert_to_tensor([[[0, 0, 0, 1], [0, 0, 0, 1]]] * 2, dtype=tf.int64) return {"input_ids": input_ids, "global_attention_mask": global_attention_mask} @unpack_inputs @@ -2547,8 +2596,8 @@ class TFLongformerForMultipleChoice(TFLongformerPreTrainedModel, TFMultipleChoic @tf.function( input_signature=[ { - "input_ids": tf.TensorSpec((None, None, None), tf.int32, name="input_ids"), - "attention_mask": tf.TensorSpec((None, None, None), tf.int32, name="attention_mask"), + "input_ids": tf.TensorSpec((None, None, None), tf.int64, name="input_ids"), + "attention_mask": tf.TensorSpec((None, None, None), tf.int64, name="attention_mask"), } ] ) diff --git a/tests/models/led/test_modeling_tf_led.py b/tests/models/led/test_modeling_tf_led.py index dfdb66606f..32ce09aaa0 100644 --- a/tests/models/led/test_modeling_tf_led.py +++ b/tests/models/led/test_modeling_tf_led.py @@ -412,7 +412,7 @@ class TFLEDModelIntegrationTest(unittest.TestCase): expected_slice = tf.convert_to_tensor( [[2.3050, 2.8279, 0.6531], [-1.8457, -0.1455, -3.5661], [-1.0186, 0.4586, -2.2043]], ) - tf.debugging.assert_near(output[:, :3, :3], expected_slice, atol=TOLERANCE) + tf.debugging.assert_near(output[:, :3, :3], expected_slice, atol=1e-3) def test_inference_with_head(self): model = TFLEDForConditionalGeneration.from_pretrained("allenai/led-base-16384") @@ -428,4 +428,4 @@ class TFLEDModelIntegrationTest(unittest.TestCase): expected_slice = tf.convert_to_tensor( [[33.6507, 6.4572, 16.8089], [5.8739, -2.4238, 11.2902], [-3.2139, -4.3149, 4.2783]], ) - tf.debugging.assert_near(output[:, :3, :3], expected_slice, atol=TOLERANCE) + tf.debugging.assert_near(output[:, :3, :3], expected_slice, atol=1e-3, rtol=1e-3) diff --git a/tests/models/longformer/test_modeling_tf_longformer.py b/tests/models/longformer/test_modeling_tf_longformer.py index cc62bb6caf..60a8ce01f4 100644 --- a/tests/models/longformer/test_modeling_tf_longformer.py +++ b/tests/models/longformer/test_modeling_tf_longformer.py @@ -115,7 +115,7 @@ class TFLongformerModelTester: ): model = TFLongformerModel(config=config) - attention_mask = tf.ones(input_ids.shape, dtype=tf.dtypes.int32) + attention_mask = tf.ones(input_ids.shape, dtype=tf.int64) output_with_mask = model(input_ids, attention_mask=attention_mask)[0] output_without_mask = model(input_ids)[0] tf.debugging.assert_near(output_with_mask[0, 0, :5], output_without_mask[0, 0, :5], rtol=1e-4) @@ -403,26 +403,24 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): # first row => [0.4983, 2.6918, -0.0071, 1.0492, 0.0000, 0.0000, 0.0000] tf.debugging.assert_near(padded_hidden_states[0, 0, 0, :4], chunked_hidden_states[0, 0, 0], rtol=1e-3) - tf.debugging.assert_near(padded_hidden_states[0, 0, 0, 4:], tf.zeros((3,), dtype=tf.dtypes.float32), rtol=1e-3) + tf.debugging.assert_near(padded_hidden_states[0, 0, 0, 4:], tf.zeros((3,), dtype=tf.float32), rtol=1e-3) # last row => [0.0000, 0.0000, 0.0000, 2.0514, -1.1600, 0.5372, 0.2629] tf.debugging.assert_near(padded_hidden_states[0, 0, -1, 3:], chunked_hidden_states[0, 0, -1], rtol=1e-3) - tf.debugging.assert_near( - padded_hidden_states[0, 0, -1, :3], tf.zeros((3,), dtype=tf.dtypes.float32), rtol=1e-3 - ) + tf.debugging.assert_near(padded_hidden_states[0, 0, -1, :3], tf.zeros((3,), dtype=tf.float32), rtol=1e-3) def test_pad_and_transpose_last_two_dims(self): hidden_states = self._get_hidden_states() self.assertEqual(shape_list(hidden_states), [1, 4, 8]) # pad along seq length dim - paddings = tf.constant([[0, 0], [0, 0], [0, 1], [0, 0]], dtype=tf.dtypes.int32) + paddings = tf.constant([[0, 0], [0, 0], [0, 1], [0, 0]], dtype=tf.int64) hidden_states = TFLongformerSelfAttention._chunk(hidden_states, window_overlap=2) padded_hidden_states = TFLongformerSelfAttention._pad_and_transpose_last_two_dims(hidden_states, paddings) self.assertTrue(shape_list(padded_hidden_states) == [1, 1, 8, 5]) - expected_added_dim = tf.zeros((5,), dtype=tf.dtypes.float32) + expected_added_dim = tf.zeros((5,), dtype=tf.float32) tf.debugging.assert_near(expected_added_dim, padded_hidden_states[0, 0, -1, :], rtol=1e-6) tf.debugging.assert_near( hidden_states[0, 0, -1, :], tf.reshape(padded_hidden_states, (1, -1))[0, 24:32], rtol=1e-6 @@ -441,10 +439,10 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): hid_states_3 = TFLongformerSelfAttention._mask_invalid_locations(hidden_states[:, :, :, :3], 2) hid_states_4 = TFLongformerSelfAttention._mask_invalid_locations(hidden_states[:, :, 2:, :], 2) - self.assertTrue(tf.math.reduce_sum(tf.cast(tf.math.is_inf(hid_states_1), tf.dtypes.int32)) == 8) - self.assertTrue(tf.math.reduce_sum(tf.cast(tf.math.is_inf(hid_states_2), tf.dtypes.int32)) == 24) - self.assertTrue(tf.math.reduce_sum(tf.cast(tf.math.is_inf(hid_states_3), tf.dtypes.int32)) == 24) - self.assertTrue(tf.math.reduce_sum(tf.cast(tf.math.is_inf(hid_states_4), tf.dtypes.int32)) == 12) + self.assertTrue(tf.math.reduce_sum(tf.cast(tf.math.is_inf(hid_states_1), tf.int64)) == 8) + self.assertTrue(tf.math.reduce_sum(tf.cast(tf.math.is_inf(hid_states_2), tf.int64)) == 24) + self.assertTrue(tf.math.reduce_sum(tf.cast(tf.math.is_inf(hid_states_3), tf.int64)) == 24) + self.assertTrue(tf.math.reduce_sum(tf.cast(tf.math.is_inf(hid_states_4), tf.int64)) == 12) def test_chunk(self): hidden_states = self._get_hidden_states() @@ -456,12 +454,14 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): chunked_hidden_states = TFLongformerSelfAttention._chunk(hidden_states, window_overlap=2) # expected slices across chunk and seq length dim - expected_slice_along_seq_length = tf.convert_to_tensor([0.4983, -0.7584, -1.6944], dtype=tf.dtypes.float32) - expected_slice_along_chunk = tf.convert_to_tensor([0.4983, -1.8348, -0.7584, 2.0514], dtype=tf.dtypes.float32) + expected_slice_along_seq_length = tf.convert_to_tensor([0.4983, -0.7584, -1.6944], dtype=tf.float32) + expected_slice_along_chunk = tf.convert_to_tensor([0.4983, -1.8348, -0.7584, 2.0514], dtype=tf.float32) self.assertTrue(shape_list(chunked_hidden_states) == [1, 3, 4, 4]) - tf.debugging.assert_near(chunked_hidden_states[0, :, 0, 0], expected_slice_along_seq_length, rtol=1e-3) - tf.debugging.assert_near(chunked_hidden_states[0, 0, :, 0], expected_slice_along_chunk, rtol=1e-3) + tf.debugging.assert_near( + chunked_hidden_states[0, :, 0, 0], expected_slice_along_seq_length, rtol=1e-3, atol=1e-4 + ) + tf.debugging.assert_near(chunked_hidden_states[0, 0, :, 0], expected_slice_along_chunk, rtol=1e-3, atol=1e-4) def test_layer_local_attn(self): model = TFLongformerModel.from_pretrained("patrickvonplaten/longformer-random-tiny") @@ -469,7 +469,7 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): hidden_states = self._get_hidden_states() batch_size, seq_length, hidden_size = hidden_states.shape - attention_mask = tf.zeros((batch_size, seq_length), dtype=tf.dtypes.float32) + attention_mask = tf.zeros((batch_size, seq_length), dtype=tf.float32) is_index_global_attn = tf.math.greater(attention_mask, 1) is_global_attn = tf.math.reduce_any(is_index_global_attn) @@ -483,11 +483,11 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): )[0] expected_slice = tf.convert_to_tensor( - [0.00188, 0.012196, -0.017051, -0.025571, -0.02996, 0.017297, -0.011521, 0.004848], dtype=tf.dtypes.float32 + [0.00188, 0.012196, -0.017051, -0.025571, -0.02996, 0.017297, -0.011521, 0.004848], dtype=tf.float32 ) self.assertEqual(output_hidden_states.shape, (1, 4, 8)) - tf.debugging.assert_near(output_hidden_states[0, 1], expected_slice, rtol=1e-3) + tf.debugging.assert_near(output_hidden_states[0, 1], expected_slice, rtol=1e-3, atol=1e-4) def test_layer_global_attn(self): model = TFLongformerModel.from_pretrained("patrickvonplaten/longformer-random-tiny") @@ -498,8 +498,8 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): batch_size, seq_length, hidden_size = hidden_states.shape # create attn mask - attention_mask_1 = tf.zeros((1, 1, 1, seq_length), dtype=tf.dtypes.float32) - attention_mask_2 = tf.zeros((1, 1, 1, seq_length), dtype=tf.dtypes.float32) + attention_mask_1 = tf.zeros((1, 1, 1, seq_length), dtype=tf.float32) + attention_mask_2 = tf.zeros((1, 1, 1, seq_length), dtype=tf.float32) attention_mask_1 = tf.where(tf.range(4)[None, :, None, None] > 1, 10000.0, attention_mask_1) attention_mask_1 = tf.where(tf.range(4)[None, :, None, None] > 2, -10000.0, attention_mask_1) @@ -525,15 +525,15 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): self.assertEqual(output_hidden_states.shape, (2, 4, 8)) expected_slice_0 = tf.convert_to_tensor( - [-0.06508, -0.039306, 0.030934, -0.03417, -0.00656, -0.01553, -0.02088, -0.04938], dtype=tf.dtypes.float32 + [-0.06508, -0.039306, 0.030934, -0.03417, -0.00656, -0.01553, -0.02088, -0.04938], dtype=tf.float32 ) expected_slice_1 = tf.convert_to_tensor( - [-0.04055, -0.038399, 0.0396, -0.03735, -0.03415, 0.01357, 0.00145, -0.05709], dtype=tf.dtypes.float32 + [-0.04055, -0.038399, 0.0396, -0.03735, -0.03415, 0.01357, 0.00145, -0.05709], dtype=tf.float32 ) - tf.debugging.assert_near(output_hidden_states[0, 2], expected_slice_0, rtol=1e-3) - tf.debugging.assert_near(output_hidden_states[1, -2], expected_slice_1, rtol=1e-3) + tf.debugging.assert_near(output_hidden_states[0, 2], expected_slice_0, rtol=1e-3, atol=1e-4) + tf.debugging.assert_near(output_hidden_states[1, -2], expected_slice_1, rtol=1e-3, atol=1e-4) def test_layer_attn_probs(self): model = TFLongformerModel.from_pretrained("patrickvonplaten/longformer-random-tiny") @@ -542,8 +542,8 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): batch_size, seq_length, hidden_size = hidden_states.shape # create attn mask - attention_mask_1 = tf.zeros((1, 1, 1, seq_length), dtype=tf.dtypes.float32) - attention_mask_2 = tf.zeros((1, 1, 1, seq_length), dtype=tf.dtypes.float32) + attention_mask_1 = tf.zeros((1, 1, 1, seq_length), dtype=tf.float32) + attention_mask_2 = tf.zeros((1, 1, 1, seq_length), dtype=tf.float32) attention_mask_1 = tf.where(tf.range(4)[None, :, None, None] > 1, 10000.0, attention_mask_1) attention_mask_1 = tf.where(tf.range(4)[None, :, None, None] > 2, -10000.0, attention_mask_1) @@ -584,18 +584,16 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): tf.debugging.assert_near( local_attentions[0, 0, 0, :], - tf.convert_to_tensor( - [0.3328, 0.0000, 0.0000, 0.0000, 0.0000, 0.3355, 0.3318, 0.0000], dtype=tf.dtypes.float32 - ), + tf.convert_to_tensor([0.3328, 0.0000, 0.0000, 0.0000, 0.0000, 0.3355, 0.3318, 0.0000], dtype=tf.float32), rtol=1e-3, + atol=1e-4, ) tf.debugging.assert_near( local_attentions[1, 0, 0, :], - tf.convert_to_tensor( - [0.2492, 0.2502, 0.2502, 0.0000, 0.0000, 0.2505, 0.0000, 0.0000], dtype=tf.dtypes.float32 - ), + tf.convert_to_tensor([0.2492, 0.2502, 0.2502, 0.0000, 0.0000, 0.2505, 0.0000, 0.0000], dtype=tf.float32), rtol=1e-3, + atol=1e-4, ) # All the global attention weights must sum to 1. @@ -603,13 +601,15 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): tf.debugging.assert_near( global_attentions[0, 0, 1, :], - tf.convert_to_tensor([0.2500, 0.2500, 0.2500, 0.2500], dtype=tf.dtypes.float32), + tf.convert_to_tensor([0.2500, 0.2500, 0.2500, 0.2500], dtype=tf.float32), rtol=1e-3, + atol=1e-4, ) tf.debugging.assert_near( global_attentions[1, 0, 0, :], - tf.convert_to_tensor([0.2497, 0.2500, 0.2499, 0.2504], dtype=tf.dtypes.float32), + tf.convert_to_tensor([0.2497, 0.2500, 0.2499, 0.2504], dtype=tf.float32), rtol=1e-3, + atol=1e-4, ) @slow @@ -617,31 +617,31 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): model = TFLongformerModel.from_pretrained("allenai/longformer-base-4096") # 'Hello world!' - input_ids = tf.convert_to_tensor([[0, 20920, 232, 328, 1437, 2]], dtype=tf.dtypes.int32) - attention_mask = tf.ones(shape_list(input_ids), dtype=tf.dtypes.int32) + input_ids = tf.convert_to_tensor([[0, 20920, 232, 328, 1437, 2]], dtype=tf.int64) + attention_mask = tf.ones(shape_list(input_ids), dtype=tf.int64) output = model(input_ids, attention_mask=attention_mask)[0] output_without_mask = model(input_ids)[0] - expected_output_slice = tf.convert_to_tensor( - [0.0549, 0.1087, -0.1119, -0.0368, 0.0250], dtype=tf.dtypes.float32 - ) + expected_output_slice = tf.convert_to_tensor([0.0549, 0.1087, -0.1119, -0.0368, 0.0250], dtype=tf.float32) - tf.debugging.assert_near(output[0, 0, -5:], expected_output_slice, rtol=1e-3) - tf.debugging.assert_near(output_without_mask[0, 0, -5:], expected_output_slice, rtol=1e-3) + tf.debugging.assert_near(output[0, 0, -5:], expected_output_slice, rtol=1e-3, atol=1e-4) + tf.debugging.assert_near(output_without_mask[0, 0, -5:], expected_output_slice, rtol=1e-3, atol=1e-4) @slow def test_inference_no_head_long(self): model = TFLongformerModel.from_pretrained("allenai/longformer-base-4096") # 'Hello world! ' repeated 1000 times - input_ids = tf.convert_to_tensor([[0] + [20920, 232, 328, 1437] * 1000 + [2]], dtype=tf.dtypes.int32) + input_ids = tf.convert_to_tensor([[0] + [20920, 232, 328, 1437] * 1000 + [2]], dtype=tf.int64) - attention_mask = tf.ones(shape_list(input_ids), dtype=tf.dtypes.int32) - global_attention_mask = tf.zeros(shape_list(input_ids), dtype=tf.dtypes.int32) + attention_mask = tf.ones(shape_list(input_ids), dtype=tf.int64) + global_attention_mask = tf.zeros(shape_list(input_ids), dtype=tf.int64) # Set global attention on a few random positions global_attention_mask = tf.tensor_scatter_nd_update( - global_attention_mask, tf.constant([[0, 1], [0, 4], [0, 21]]), tf.constant([1, 1, 1]) + global_attention_mask, + tf.constant([[0, 1], [0, 4], [0, 21]], dtype=tf.int64), + tf.constant([1, 1, 1], dtype=tf.int64), ) output = model(input_ids, attention_mask=attention_mask, global_attention_mask=global_attention_mask)[0] @@ -650,15 +650,15 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): expected_output_mean = tf.constant(0.024267) # assert close - tf.debugging.assert_near(tf.reduce_sum(output), expected_output_sum, rtol=1e-4) - tf.debugging.assert_near(tf.reduce_mean(output), expected_output_mean, rtol=1e-4) + tf.debugging.assert_near(tf.reduce_sum(output), expected_output_sum, rtol=1e-4, atol=1e-4) + tf.debugging.assert_near(tf.reduce_mean(output), expected_output_mean, rtol=1e-4, atol=1e-4) @slow def test_inference_masked_lm_long(self): model = TFLongformerForMaskedLM.from_pretrained("allenai/longformer-base-4096") # 'Hello world! ' repeated 1000 times - input_ids = tf.convert_to_tensor([[0] + [20920, 232, 328, 1437] * 1000 + [2]], dtype=tf.dtypes.int32) + input_ids = tf.convert_to_tensor([[0] + [20920, 232, 328, 1437] * 1000 + [2]], dtype=tf.int64) output = model(input_ids, labels=input_ids) loss = output.loss @@ -669,9 +669,13 @@ class TFLongformerModelIntegrationTest(unittest.TestCase): expected_prediction_scores_mean = tf.constant(-3.03477) # assert close - tf.debugging.assert_near(tf.reduce_mean(loss), expected_loss, rtol=1e-4) - tf.debugging.assert_near(tf.reduce_sum(prediction_scores), expected_prediction_scores_sum, rtol=1e-4) - tf.debugging.assert_near(tf.reduce_mean(prediction_scores), expected_prediction_scores_mean, rtol=1e-4) + tf.debugging.assert_near(tf.reduce_mean(loss), expected_loss, rtol=1e-4, atol=1e-4) + tf.debugging.assert_near( + tf.reduce_sum(prediction_scores), expected_prediction_scores_sum, rtol=1e-4, atol=1e-4 + ) + tf.debugging.assert_near( + tf.reduce_mean(prediction_scores), expected_prediction_scores_mean, rtol=1e-4, atol=1e-4 + ) @slow def test_inference_masked_lm(self):