wehub-resource-sync
686 lines
27 KiB
Python
686 lines
27 KiB
Python
# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
|
|
# Copyright 2020 The HuggingFace Team. All rights reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
from __future__ import annotations
|
|
|
|
import copy
|
|
import math
|
|
import random
|
|
import tempfile
|
|
import unittest
|
|
|
|
import numpy as np
|
|
import paddle
|
|
import pytest
|
|
from parameterized import parameterized, parameterized_class
|
|
|
|
from paddlenlp.transformers import (
|
|
BloomConfig,
|
|
BloomForCausalLM,
|
|
BloomForSequenceClassification,
|
|
BloomForTokenClassification,
|
|
BloomModel,
|
|
BloomTokenizer,
|
|
)
|
|
from paddlenlp.transformers.bloom.modeling import BloomForGeneration
|
|
from tests.testing_utils import PaddleNLPModelTest, require_package, slow
|
|
from tests.transformers.test_generation_utils import GenerationTesterMixin
|
|
from tests.transformers.test_modeling_common import ( # GenerationD2STestMixin,
|
|
ModelTesterMixin,
|
|
floats_tensor,
|
|
ids_tensor,
|
|
random_attention_mask,
|
|
)
|
|
|
|
|
|
class BloomModelTester:
|
|
def __init__(
|
|
self,
|
|
parent,
|
|
batch_size=2,
|
|
seq_length=20,
|
|
is_training=False,
|
|
use_input_mask=True,
|
|
vocab_size=100,
|
|
hidden_size=32,
|
|
n_layer=2,
|
|
n_head=8,
|
|
masked_softmax_fusion=True,
|
|
layer_norm_epsilon=1e-5,
|
|
initializer_range=0.02,
|
|
use_cache=False,
|
|
bos_token_id=1,
|
|
eos_token_id=2,
|
|
apply_residual_connection_post_layernorm=False,
|
|
hidden_dropout=0.0,
|
|
attention_dropout=0.0,
|
|
attention_softmax_in_fp32=True,
|
|
pretraining_tp=1, # TP rank used when training with megatron
|
|
type_sequence_label_size=2,
|
|
num_labels=3,
|
|
num_choices=4,
|
|
scope=None,
|
|
):
|
|
self.parent = parent
|
|
self.batch_size = batch_size
|
|
self.seq_length = seq_length
|
|
self.is_training = is_training
|
|
self.use_input_mask = use_input_mask
|
|
self.vocab_size = vocab_size
|
|
self.hidden_size = hidden_size
|
|
self.num_attention_heads = self.n_head = n_head
|
|
self.num_hidden_layers = self.n_layer = n_layer
|
|
self.n_head = n_head
|
|
self.masked_softmax_fusion = masked_softmax_fusion
|
|
self.layer_norm_epsilon = layer_norm_epsilon
|
|
self.initializer_range = initializer_range
|
|
self.use_cache = use_cache
|
|
self.bos_token_id = bos_token_id
|
|
self.eos_token_id = eos_token_id
|
|
self.apply_residual_connection_post_layernorm = apply_residual_connection_post_layernorm
|
|
self.hidden_dropout = hidden_dropout
|
|
self.attention_dropout = attention_dropout
|
|
self.attention_softmax_in_fp32 = attention_softmax_in_fp32
|
|
self.pretraining_tp = pretraining_tp
|
|
self.num_labels = num_labels
|
|
self.num_choices = num_choices
|
|
self.type_sequence_label_size = type_sequence_label_size
|
|
self.scope = None
|
|
self.bos_token_id = 1
|
|
self.eos_token_id = 2
|
|
self.pad_token_id = 3
|
|
|
|
def prepare_config_and_inputs(self):
|
|
input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size, dtype="int64")
|
|
|
|
input_mask = None
|
|
if self.use_input_mask:
|
|
input_mask = random_attention_mask([self.batch_size, self.seq_length], dtype="int64")
|
|
|
|
sequence_labels = None
|
|
token_labels = None
|
|
choice_labels = None
|
|
|
|
if self.parent.use_labels:
|
|
sequence_labels = ids_tensor([self.batch_size], self.type_sequence_label_size, dtype="int64")
|
|
token_labels = ids_tensor([self.batch_size, self.seq_length], self.num_labels, dtype="int64")
|
|
choice_labels = ids_tensor([self.batch_size], self.num_choices, dtype="int64")
|
|
|
|
config = self.get_config()
|
|
|
|
return (
|
|
config,
|
|
input_ids,
|
|
input_mask,
|
|
sequence_labels,
|
|
token_labels,
|
|
choice_labels,
|
|
)
|
|
|
|
def get_config(self):
|
|
return BloomConfig(
|
|
vocab_size=self.vocab_size,
|
|
hidden_size=self.hidden_size,
|
|
n_layer=self.n_layer,
|
|
n_head=self.n_head,
|
|
masked_softmax_fusion=self.masked_softmax_fusion,
|
|
layer_norm_epsilon=self.layer_norm_epsilon,
|
|
initializer_range=self.initializer_range,
|
|
use_cache=self.use_cache,
|
|
bos_token_id=self.bos_token_id,
|
|
eos_token_id=self.eos_token_id,
|
|
apply_residual_connection_post_layernorm=self.apply_residual_connection_post_layernorm,
|
|
hidden_dropout=self.hidden_dropout,
|
|
attention_dropout=self.attention_dropout,
|
|
attention_softmax_in_fp32=self.attention_softmax_in_fp32,
|
|
pretraining_tp=self.pretraining_tp,
|
|
num_labels=self.num_labels,
|
|
num_choices=self.num_choices,
|
|
)
|
|
|
|
def prepare_config_and_inputs_for_decoder(self):
|
|
(
|
|
config,
|
|
input_ids,
|
|
input_mask,
|
|
sequence_labels,
|
|
token_labels,
|
|
choice_labels,
|
|
) = self.prepare_config_and_inputs()
|
|
|
|
encoder_hidden_states = floats_tensor([self.batch_size, self.seq_length, self.hidden_size])
|
|
encoder_attention_mask = paddle.cast(
|
|
ids_tensor([self.batch_size, self.seq_length], vocab_size=2), dtype="float32"
|
|
)
|
|
|
|
return (
|
|
config,
|
|
input_ids,
|
|
input_mask,
|
|
sequence_labels,
|
|
token_labels,
|
|
choice_labels,
|
|
encoder_hidden_states,
|
|
encoder_attention_mask,
|
|
)
|
|
|
|
def create_and_check_gpt_model(self, config, input_ids, input_mask, *args):
|
|
model = BloomModel(config)
|
|
model.eval()
|
|
|
|
result = model(input_ids, use_cache=True, return_dict=self.parent.return_dict)
|
|
result = model(input_ids, use_cache=True, return_dict=self.parent.return_dict)
|
|
result = model(input_ids, use_cache=True, return_dict=self.parent.return_dict)
|
|
|
|
self.parent.assertEqual(result[0].shape, [self.batch_size, self.seq_length, self.hidden_size])
|
|
self.parent.assertEqual(len(result[1]), config["n_layer"])
|
|
|
|
def create_and_check_gpt_model_past(self, config, input_ids, input_mask, *args):
|
|
model = BloomModel(config)
|
|
model.eval()
|
|
|
|
# first forward pass
|
|
outputs = model(input_ids, use_cache=False, return_dict=self.parent.return_dict)
|
|
outputs_use_cache_conf = model(input_ids, use_cache=True, return_dict=self.parent.return_dict)
|
|
|
|
self.parent.assertTrue(len(outputs) + 1 == len(outputs_use_cache_conf))
|
|
|
|
output, past = outputs_use_cache_conf[:2]
|
|
|
|
# create hypothetical next token and extent to next_input_ids
|
|
next_tokens = ids_tensor((self.batch_size, 1), config["vocab_size"], dtype="int64")
|
|
|
|
# append to next input_ids
|
|
next_input_ids = paddle.concat([input_ids, next_tokens], axis=-1)
|
|
|
|
output_from_no_past = model(next_input_ids, return_dict=self.parent.return_dict)[0]
|
|
output_from_past = model(next_tokens, use_cache=True, cache=past, return_dict=self.parent.return_dict)[0]
|
|
|
|
# select random slice
|
|
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1]).item()
|
|
output_from_no_past_slice = output_from_no_past[:, -1, random_slice_idx].detach()
|
|
output_from_past_slice = output_from_past[:, 0, random_slice_idx].detach()
|
|
|
|
# test that outputs are equal for slice
|
|
self.parent.assertTrue(paddle.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-3))
|
|
|
|
def create_and_check_gpt_model_attention_mask_past(self, config, input_ids, input_mask, *args):
|
|
model = BloomModel(config)
|
|
model.eval()
|
|
|
|
# create attention mask
|
|
attn_mask = paddle.ones(input_ids.shape, dtype="float32")
|
|
half_seq_length = self.seq_length // 2
|
|
attn_mask[:, half_seq_length:] = 0
|
|
|
|
# first forward pass
|
|
output, past = model(input_ids, attention_mask=attn_mask, use_cache=True, return_dict=self.parent.return_dict)[
|
|
:2
|
|
]
|
|
|
|
# create hypothetical next token and extent to next_input_ids
|
|
next_tokens = ids_tensor((self.batch_size, 1), config["vocab_size"], dtype="int64")
|
|
|
|
# change a random masked slice from input_ids
|
|
random_seq_idx_to_change = ids_tensor((1,), half_seq_length, dtype="int64").item() + 1
|
|
random_other_next_tokens = ids_tensor((self.batch_size, 1), config["vocab_size"], dtype="int64").squeeze(-1)
|
|
input_ids[:, -random_seq_idx_to_change] = random_other_next_tokens
|
|
|
|
# append to next input_ids and attn_mask
|
|
next_input_ids = paddle.concat([input_ids, next_tokens], axis=-1)
|
|
attn_mask = paddle.concat(
|
|
[attn_mask, paddle.ones((attn_mask.shape[0], 1), dtype="float32")],
|
|
axis=1,
|
|
)
|
|
|
|
# get two different outputs
|
|
output_from_no_past = model(next_input_ids, attention_mask=attn_mask, return_dict=self.parent.return_dict)[0]
|
|
output_from_past = model(
|
|
next_tokens, cache=past, use_cache=True, attention_mask=attn_mask, return_dict=self.parent.return_dict
|
|
)[0]
|
|
|
|
# select random slice
|
|
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1], dtype="int64").item()
|
|
output_from_no_past_slice = output_from_no_past[:, -1, random_slice_idx]
|
|
output_from_past_slice = output_from_past[:, 0, random_slice_idx]
|
|
|
|
# test that outputs are equal for slice
|
|
self.parent.assertTrue(paddle.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-3))
|
|
|
|
def create_and_check_gpt_model_past_large_inputs(self, config, input_ids, input_mask, *args):
|
|
model = BloomModel(config)
|
|
model.eval()
|
|
|
|
# first forward pass
|
|
outputs = model(input_ids, attention_mask=input_mask, use_cache=True, return_dict=self.parent.return_dict)
|
|
|
|
output, past = outputs[:2]
|
|
|
|
# create hypothetical next token and extent to next_input_ids
|
|
next_tokens = ids_tensor((self.batch_size, 3), config["vocab_size"], dtype="int64")
|
|
next_mask = ids_tensor((self.batch_size, 3), vocab_size=2, dtype="int64")
|
|
|
|
# append to next input_ids
|
|
next_input_ids = paddle.concat([input_ids, next_tokens], axis=-1)
|
|
next_attention_mask = paddle.concat([input_mask, next_mask], axis=-1)
|
|
|
|
output_from_no_past = model(
|
|
next_input_ids, attention_mask=next_attention_mask, return_dict=self.parent.return_dict
|
|
)[0]
|
|
output_from_past = model(
|
|
next_tokens,
|
|
attention_mask=next_attention_mask,
|
|
cache=past,
|
|
use_cache=True,
|
|
return_dict=self.parent.return_dict,
|
|
)[0]
|
|
self.parent.assertTrue(output_from_past.shape[1] == next_tokens.shape[1])
|
|
|
|
# select random slice
|
|
random_slice_idx = ids_tensor((1,), output_from_past.shape[-1], dtype="int64").item()
|
|
output_from_no_past_slice = output_from_no_past[:, -3:, random_slice_idx].detach()
|
|
output_from_past_slice = output_from_past[:, :, random_slice_idx].detach()
|
|
|
|
# test that outputs are equal for slice
|
|
self.parent.assertTrue(paddle.allclose(output_from_past_slice, output_from_no_past_slice, atol=1e-3))
|
|
|
|
def create_and_check_lm_head_model(self, config, input_ids, input_mask, *args):
|
|
model = BloomForCausalLM(config)
|
|
model.eval()
|
|
|
|
result = model(
|
|
input_ids,
|
|
use_cache=True,
|
|
labels=input_ids if self.parent.use_labels else None,
|
|
return_dict=self.parent.return_dict,
|
|
)
|
|
if self.parent.use_labels:
|
|
self.parent.assertIsInstance(result[0].item(), float)
|
|
self.parent.assertEqual(result[1].shape, [self.batch_size, self.seq_length, self.vocab_size])
|
|
else:
|
|
self.parent.assertEqual(result[0].shape, [self.batch_size, self.seq_length, self.vocab_size])
|
|
|
|
def create_and_check_forward_and_backwards(self, config, input_ids, input_mask, *args):
|
|
model = BloomForCausalLM(config)
|
|
|
|
if self.parent.use_labels:
|
|
loss, logits = model(input_ids, labels=input_ids, return_dict=self.parent.return_dict)
|
|
self.parent.assertEqual(loss.shape, [1])
|
|
self.parent.assertEqual(logits.shape, [self.batch_size, self.seq_length, self.vocab_size])
|
|
loss.backward()
|
|
|
|
def create_and_check_gpt_for_sequence_classification(self, config, input_ids, input_mask, sequence_labels, *args):
|
|
config.num_labels = self.num_labels
|
|
model = BloomForSequenceClassification(config)
|
|
model.eval()
|
|
result = model(
|
|
input_ids,
|
|
attention_mask=input_mask,
|
|
labels=sequence_labels if self.parent.use_labels else None,
|
|
return_dict=self.parent.return_dict,
|
|
)
|
|
if self.parent.use_labels:
|
|
self.parent.assertIsInstance(result[0].item(), float)
|
|
self.parent.assertEqual(result[1].shape, [self.batch_size, self.num_labels])
|
|
else:
|
|
self.parent.assertEqual(result[0].shape, [self.batch_size, self.num_labels])
|
|
|
|
def create_and_check_gpt_for_token_classification(
|
|
self, config, input_ids, input_mask, sequence_labels, token_labels, *args
|
|
):
|
|
config.num_labels = self.num_labels
|
|
model = BloomForTokenClassification(config)
|
|
model.eval()
|
|
result = model(
|
|
input_ids,
|
|
attention_mask=input_mask,
|
|
labels=token_labels if self.parent.use_labels else None,
|
|
return_dict=self.parent.return_dict,
|
|
)
|
|
if self.parent.use_labels:
|
|
self.parent.assertIsInstance(result[0].item(), float)
|
|
self.parent.assertEqual(result[1].shape, [self.batch_size, self.seq_length, self.num_labels])
|
|
else:
|
|
self.parent.assertEqual(result[0].shape, [self.batch_size, self.seq_length, self.num_labels])
|
|
|
|
def create_and_check_gpt_weight_initialization(self, config, *args):
|
|
model = BloomModel(config)
|
|
model_std = model.config["initializer_range"] / math.sqrt(2 * model.config["n_layer"])
|
|
for key in model.state_dict().keys():
|
|
if "out_proj" in key and "weight" in key:
|
|
self.parent.assertLessEqual(abs((paddle.std(model.state_dict()[key]) - model_std).numpy()), 0.02)
|
|
self.parent.assertLessEqual(abs((paddle.mean(model.state_dict()[key]) - 0.0).numpy()), 0.01)
|
|
|
|
def create_and_check_model_attention_mask(
|
|
self, config: BloomConfig, input_ids, input_mask, sequence_labels, token_labels, choice_labels
|
|
):
|
|
model = BloomModel(config)
|
|
model.eval()
|
|
attn_mask_2d = random_attention_mask([self.batch_size, self.seq_length])
|
|
result_2d = model(input_ids, attention_mask=attn_mask_2d)[0]
|
|
batch, seq_length = input_ids.shape
|
|
causal_mask = paddle.tril(paddle.ones((batch, seq_length, seq_length), dtype=attn_mask_2d.dtype))
|
|
attn_mask_3d = causal_mask & attn_mask_2d.unsqueeze(-1)
|
|
result_3d = model(input_ids, attention_mask=attn_mask_3d)[0]
|
|
attn_mask_4d = attn_mask_3d.unsqueeze(1)
|
|
result_4d = model(input_ids, attention_mask=attn_mask_4d)[0]
|
|
result_no_attention_mask = model(input_ids, attention_mask=None)[0]
|
|
# Assert non-padding tokens have the same logits with different attention_mask shape
|
|
self.parent.assertTrue((result_2d[attn_mask_2d] == result_3d[attn_mask_2d]).all())
|
|
self.parent.assertTrue((result_2d[attn_mask_2d] == result_4d[attn_mask_2d]).all())
|
|
self.parent.assertTrue((result_2d[attn_mask_2d] == result_no_attention_mask[attn_mask_2d]).all())
|
|
|
|
def prepare_config_and_inputs_for_common(self):
|
|
config_and_inputs = self.prepare_config_and_inputs()
|
|
(
|
|
config,
|
|
input_ids,
|
|
input_mask,
|
|
sequence_labels,
|
|
token_labels,
|
|
choice_labels,
|
|
) = config_and_inputs
|
|
|
|
inputs_dict = {
|
|
"input_ids": input_ids,
|
|
}
|
|
|
|
return config, inputs_dict
|
|
|
|
def prepare_config_and_inputs_for_gpt(self):
|
|
config = self.get_config()
|
|
# excluding eos_token_id which is equal to vocab_size - 1
|
|
input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size - 1, dtype="int64")
|
|
inputs_dict = {"input_ids": input_ids}
|
|
return config, inputs_dict
|
|
|
|
|
|
@parameterized_class(
|
|
("return_dict", "use_labels"),
|
|
[[False, False], [False, True], [True, False], [True, True]],
|
|
)
|
|
class BloomModelTest(ModelTesterMixin, GenerationTesterMixin, unittest.TestCase):
|
|
base_model_class = BloomModel
|
|
use_labels = False
|
|
return_dict = False
|
|
use_test_model_name_list = False
|
|
|
|
all_model_classes = (BloomModel, BloomForCausalLM, BloomForSequenceClassification, BloomForTokenClassification)
|
|
all_generative_model_classes = {BloomForCausalLM: (BloomModel, "bloom")}
|
|
|
|
all_parallelizable_model_classes = BloomForCausalLM
|
|
test_missing_keys = False
|
|
test_tie_weights = False
|
|
test_model_parallel = True
|
|
|
|
# special case for DoubleHeads model
|
|
def _prepare_for_class(self, inputs_dict, model_class):
|
|
inputs_dict = super()._prepare_for_class(inputs_dict, model_class)
|
|
|
|
return inputs_dict
|
|
|
|
def setUp(self):
|
|
self.model_tester = BloomModelTester(self)
|
|
self.test_resize_embeddings = False
|
|
random.seed(128)
|
|
np.random.seed(128)
|
|
paddle.seed(128)
|
|
|
|
def test_gpt_model(self):
|
|
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
|
self.model_tester.create_and_check_gpt_model(*config_and_inputs)
|
|
|
|
def test_gpt_model_past(self):
|
|
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
|
self.model_tester.create_and_check_gpt_model_past(*config_and_inputs)
|
|
|
|
def test_gpt_model_att_mask_past(self):
|
|
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
|
self.model_tester.create_and_check_gpt_model_attention_mask_past(*config_and_inputs)
|
|
|
|
def test_gpt_model_past_large_inputs(self):
|
|
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
|
self.model_tester.create_and_check_gpt_model_past_large_inputs(*config_and_inputs)
|
|
|
|
def test_gpt_lm_head_model(self):
|
|
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
|
self.model_tester.create_and_check_lm_head_model(*config_and_inputs)
|
|
|
|
def test_gpt_sequence_classification_model(self):
|
|
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
|
self.model_tester.create_and_check_gpt_for_sequence_classification(*config_and_inputs)
|
|
|
|
def test_gpt_token_classification_model(self):
|
|
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
|
self.model_tester.create_and_check_gpt_for_token_classification(*config_and_inputs)
|
|
|
|
def test_gpt_weight_initialization(self):
|
|
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
|
self.model_tester.create_and_check_gpt_weight_initialization(*config_and_inputs)
|
|
|
|
def test_model_attention_mask(self):
|
|
config_and_inputs = self.model_tester.prepare_config_and_inputs()
|
|
self.model_tester.create_and_check_model_attention_mask(*config_and_inputs)
|
|
|
|
def test_inputs_embeds(self):
|
|
# NOTE: rewrite test inputs embeds for gpt model since couldn't detect eos token id from inputs_embeds
|
|
# get config for model and inputs_dict for model forward
|
|
config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_gpt()
|
|
# test all model classes
|
|
for model_class in self.all_model_classes:
|
|
model = self._make_model_instance(config, model_class)
|
|
model.eval()
|
|
|
|
inputs = copy.deepcopy(self._prepare_for_class(inputs_dict, model_class))
|
|
|
|
with paddle.no_grad():
|
|
ids_output = model(**inputs)
|
|
|
|
if not self.is_encoder_decoder:
|
|
input_ids = inputs["input_ids"]
|
|
del inputs["input_ids"]
|
|
else:
|
|
encoder_input_ids = inputs["input_ids"]
|
|
decoder_input_ids = inputs.get("decoder_input_ids", encoder_input_ids)
|
|
del inputs["input_ids"]
|
|
inputs.pop("decoder_input_ids", None)
|
|
|
|
wte = model.get_input_embeddings()
|
|
if not self.is_encoder_decoder:
|
|
inputs["inputs_embeds"] = wte(input_ids)
|
|
else:
|
|
inputs["inputs_embeds"] = wte(encoder_input_ids)
|
|
inputs["decoder_inputs_embeds"] = wte(decoder_input_ids)
|
|
|
|
with paddle.no_grad():
|
|
embeds_output = model(**inputs)
|
|
if isinstance(ids_output, tuple):
|
|
ids_output = ids_output[0]
|
|
if isinstance(embeds_output, tuple):
|
|
embeds_output = embeds_output[0]
|
|
|
|
self.assertTrue(paddle.allclose(ids_output, embeds_output, rtol=1e-4, atol=1e-4))
|
|
|
|
|
|
class BloomCompatibilityTest(unittest.TestCase):
|
|
test_model_id = "hf-internal-testing/tiny-random-BloomModel"
|
|
|
|
@classmethod
|
|
@require_package("transformers", "torch")
|
|
def setUpClass(cls) -> None:
|
|
from transformers import BloomModel
|
|
|
|
# when python application is done, `TemporaryDirectory` will be free
|
|
cls.torch_model_path = tempfile.TemporaryDirectory().name
|
|
model = BloomModel.from_pretrained(cls.test_model_id)
|
|
model.save_pretrained(cls.torch_model_path)
|
|
|
|
@parameterized.expand(
|
|
[
|
|
("BloomModel", "BloomModel"),
|
|
("BloomForSequenceClassification", "BloomForSequenceClassification"),
|
|
("BloomForTokenClassification", "BloomForTokenClassification"),
|
|
("BloomForCausalLM", "BloomForCausalLM"),
|
|
]
|
|
)
|
|
@require_package("transformers", "torch")
|
|
def test_gpt_classes_from_local_dir(self, paddle_class_name, pytorch_class_name=None):
|
|
pytorch_class_name = pytorch_class_name or paddle_class_name
|
|
with tempfile.TemporaryDirectory() as tempdir:
|
|
|
|
# 1. create common input
|
|
input_ids = np.random.randint(100, 200, [1, 20])
|
|
|
|
# 2. forward the torch model
|
|
import torch
|
|
import transformers
|
|
|
|
torch_model_class = getattr(transformers, pytorch_class_name)
|
|
torch_model = torch_model_class.from_pretrained(self.torch_model_path)
|
|
torch_model.eval()
|
|
|
|
torch_model.save_pretrained(tempdir)
|
|
torch_logit = torch_model(torch.tensor(input_ids), return_dict=False)[0]
|
|
|
|
# 3. forward the paddle model
|
|
from paddlenlp import transformers
|
|
|
|
paddle_model_class = getattr(transformers, paddle_class_name)
|
|
paddle_model = paddle_model_class.from_pretrained(tempdir, convert_from_torch=True)
|
|
paddle_model.eval()
|
|
|
|
paddle_logit = paddle_model(paddle.to_tensor(input_ids), return_dict=False)[0]
|
|
|
|
self.assertTrue(
|
|
np.allclose(
|
|
paddle_logit.detach().cpu().numpy().reshape([-1])[:9],
|
|
torch_logit.detach().cpu().numpy().reshape([-1])[:9],
|
|
atol=1e-3,
|
|
)
|
|
)
|
|
|
|
|
|
class BloomModelLanguageGenerationTest(PaddleNLPModelTest):
|
|
def _test_lm_generate_gpt_helper(
|
|
self,
|
|
verify_outputs=True,
|
|
):
|
|
model = BloomForCausalLM.from_pretrained("bigscience/bloom-560m")
|
|
model.eval()
|
|
|
|
# The dog
|
|
input_ids = paddle.to_tensor([[464, 3290]], dtype="int64")
|
|
|
|
# The dog was found in a field near the intersection of West and West Streets.\n\nThe dog
|
|
# fmt: off
|
|
expected_output_ids = [
|
|
373,
|
|
1043,
|
|
287,
|
|
257,
|
|
2214,
|
|
1474,
|
|
262,
|
|
16246,
|
|
286,
|
|
2688,
|
|
290,
|
|
2688,
|
|
27262,
|
|
13,
|
|
198,
|
|
198,
|
|
464,
|
|
3290,
|
|
]
|
|
# fmt: on
|
|
output_ids, _ = model.generate(input_ids, decode_strategy="greedy_search", max_length=18)
|
|
if verify_outputs:
|
|
self.assertListEqual(output_ids[0].tolist(), expected_output_ids)
|
|
|
|
@pytest.mark.skip("complete `generate` method in another pr")
|
|
@slow
|
|
def test_lm_generate_gpt(self):
|
|
self._test_lm_generate_gpt_helper()
|
|
|
|
@slow
|
|
def test_gpt_for_generation(self):
|
|
model_name = "bigscience/bloom-560m"
|
|
tokenizer = BloomTokenizer.from_pretrained(model_name)
|
|
|
|
config = BloomConfig.from_pretrained(model_name)
|
|
config.top_k = 1
|
|
model = BloomForGeneration.from_pretrained(model_name, config=config)
|
|
model.eval()
|
|
|
|
paddle.seed(128)
|
|
np.random.seed(128)
|
|
random.seed(128)
|
|
|
|
tokenized = tokenizer("I love you,", return_tensors="pd")
|
|
input_ids = tokenized["input_ids"]
|
|
|
|
output_ids, _ = model(
|
|
input_ids,
|
|
)
|
|
output_str = tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
|
print(output_str)
|
|
|
|
output_seq, _ = model(input_ids=input_ids)
|
|
output_seq_strs = tokenizer.batch_decode(output_seq, skip_special_tokens=True)
|
|
print(output_seq_strs)
|
|
|
|
EXPECTED_OUTPUT_STR = " baby.\nI love you, baby.\nI love you, baby.\nI love you, baby.\n"
|
|
|
|
self.assertEqual(output_seq_strs[0], EXPECTED_OUTPUT_STR)
|
|
self.assertEqual(output_str, EXPECTED_OUTPUT_STR)
|
|
|
|
@pytest.mark.skip("complete `generate` method in another pr")
|
|
@slow
|
|
def test_gpt_sample(self):
|
|
tokenizer = BloomTokenizer.from_pretrained("bigscience/bloom-560m")
|
|
model = BloomForCausalLM.from_pretrained("bigscience/bloom-560m")
|
|
model.eval()
|
|
|
|
paddle.seed(128)
|
|
np.random.seed(128)
|
|
random.seed(128)
|
|
|
|
tokenized = tokenizer("where is the capital of china: ", return_tensors="pd")
|
|
input_ids = tokenized["input_ids"]
|
|
|
|
output_ids, _ = model.generate(
|
|
input_ids,
|
|
top_k=1,
|
|
)
|
|
output_str = tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
|
print(output_str)
|
|
|
|
output_seq, _ = model.generate(
|
|
input_ids=input_ids,
|
|
top_k=1,
|
|
)
|
|
output_seq_strs = tokenizer.batch_decode(output_seq, skip_special_tokens=True)
|
|
print(output_seq_strs)
|
|
|
|
EXPECTED_OUTPUT_STR = "the result is not accurate with BloomForGeneration."
|
|
|
|
self.assertEqual(output_seq_strs[0], EXPECTED_OUTPUT_STR)
|
|
self.assertEqual(output_str, EXPECTED_OUTPUT_STR)
|
|
|
|
|
|
# class BloomGenerationD2STest(GenerationD2STestMixin, unittest.TestCase):
|
|
# max_length = 100
|
|
# internal_testing_model = "__internal_testing__/tiny-random-bloom"
|