wehub-resource-sync
77 lines
3.1 KiB
Python
77 lines
3.1 KiB
Python
# Copyright (c) 2023 PaddlePaddle Authors. 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.
|
|
|
|
# ****** Unitest not yet support paddlepaddle>=2.5.2 ******
|
|
# import unittest
|
|
# from tempfile import TemporaryDirectory
|
|
|
|
# import numpy as np
|
|
# import paddle
|
|
# from paddle.nn.quant import weight_quantize
|
|
# from parameterized import parameterized
|
|
|
|
# from paddlenlp.transformers import AutoModel
|
|
# from paddlenlp.utils.quantization import QuantizationLinear
|
|
|
|
|
|
# class TestQuantizationLinear(unittest.TestCase):
|
|
# @parameterized.expand([("weight_only_int8",), ("weight_only_int4",), ("llm.int8",)])
|
|
# def test_forward(self, quant_algo):
|
|
# qlinear = QuantizationLinear(in_features=64, out_features=64, quant_algo=quant_algo, dtype="float16")
|
|
# x = paddle.randn([2, 4, 64], "float16")
|
|
# weight = paddle.randn([64, 64], "float16")
|
|
# quant_weight, quant_scale = weight_quantize(weight, quant_algo)
|
|
# qlinear.quant_weight.set_value(quant_weight)
|
|
# qlinear.quant_scale.set_value(quant_scale)
|
|
|
|
# output = qlinear(x)
|
|
# self.assertEqual(output.shape, [2, 4, 64])
|
|
|
|
|
|
# class TestQuantizationModel(unittest.TestCase):
|
|
# @classmethod
|
|
# def setUpClass(cls):
|
|
# cls.model1 = AutoModel.from_pretrained(
|
|
# "__internal_testing__/test-tiny-random-llama",
|
|
# quantization_config={"quant_algo": "weight_only_int8"},
|
|
# dtype="float16",
|
|
# )
|
|
# cls.model2 = AutoModel.from_pretrained(
|
|
# "__internal_testing__/test-tiny-random-llama",
|
|
# quantization_config={"quant_algo": "weight_only_int4"},
|
|
# dtype="float16",
|
|
# )
|
|
# cls.model3 = AutoModel.from_pretrained(
|
|
# "__internal_testing__/test-tiny-random-llama",
|
|
# quantization_config={"quant_algo": "llm.int8"},
|
|
# dtype="float16",
|
|
# )
|
|
# cls.model1.eval()
|
|
# cls.model2.eval()
|
|
# cls.model3.eval()
|
|
|
|
# def test_forward(self):
|
|
# input_ids = paddle.to_tensor(np.random.randint(0, 128, [1, 20]))
|
|
# for model in [self.model1, self.model2, self.model3]:
|
|
# output = model(input_ids)
|
|
# self.assertEqual(output[0].shape, [1, 20, 64])
|
|
|
|
# def test_save_pretrained(self):
|
|
# input_ids = paddle.to_tensor(np.random.randint(0, 128, [1, 20]))
|
|
# for model in [self.model1, self.model2, self.model3]:
|
|
# with TemporaryDirectory() as tempdir:
|
|
# model.save_pretrained(tempdir)
|
|
# model_save = AutoModel.from_pretrained(tempdir)
|
|
# self.assertTrue(paddle.allclose(model(input_ids)[0], model_save(input_ids)[0]))
|