mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
b777bd6475
* Update TensorRT-LLM --------- Co-authored-by: RunningLeon <mnsheng@yeah.net> Co-authored-by: Tlntin <TlntinDeng01@Gmail.com> Co-authored-by: ZHENG, Zhen <zhengzhen.z@qq.com> Co-authored-by: Pham Van Ngoan <ngoanpham1196@gmail.com> Co-authored-by: Nathan Price <nathan@abridge.com> Co-authored-by: Tushar Goel <tushar.goel.ml@gmail.com> Co-authored-by: Mati <132419219+matichon-vultureprime@users.noreply.github.com>
69 lines
2.4 KiB
Python
69 lines
2.4 KiB
Python
# SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
|
# SPDX-License-Identifier: Apache-2.0
|
|
#
|
|
# 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.
|
|
import os
|
|
import sys
|
|
import unittest
|
|
|
|
import torch
|
|
from parameterized import parameterized
|
|
|
|
import tensorrt_llm
|
|
|
|
sys.path.append(os.path.join(os.path.dirname(__file__), '..'))
|
|
from utils.util import create_session, run_session, unittest_name_func
|
|
|
|
|
|
class TestTopK(unittest.TestCase):
|
|
|
|
def setUp(self):
|
|
tensorrt_llm.logger.set_level('error')
|
|
|
|
@parameterized.expand([
|
|
((3, 5), 1, 1, True),
|
|
((3, 4, 6), 2, 0, True),
|
|
((3, 5), 1, 1, False),
|
|
((3, 4, 6), 2, 0, False),
|
|
],
|
|
name_func=unittest_name_func)
|
|
def test_topk(self, input_shape, k, d, largest):
|
|
value_dtype = 'float32'
|
|
indices_dtype = 'int32'
|
|
# construct trt network
|
|
builder = tensorrt_llm.Builder()
|
|
network = builder.create_network()
|
|
input_data = torch.rand(*input_shape,
|
|
dtype=torch.float32,
|
|
device="cuda")
|
|
with tensorrt_llm.net_guard(network):
|
|
|
|
m = tensorrt_llm.functional.constant(input_data.cpu().numpy())
|
|
topk_values, topk_indices = tensorrt_llm.functional.topk(
|
|
m, k, d, largest=largest)
|
|
topk_values.mark_output('output_values', value_dtype)
|
|
topk_indices.mark_output('topk_indices', indices_dtype)
|
|
|
|
# trt run
|
|
session = create_session(builder, network)
|
|
inputs = {}
|
|
outputs = run_session(session, inputs)
|
|
|
|
# pytorch run
|
|
values, indices = torch.topk(input_data, k, dim=d, largest=largest)
|
|
|
|
# compare diff
|
|
torch.testing.assert_close(values, outputs['output_values'])
|
|
# dtype does not match
|
|
torch.testing.assert_close(indices.int(), outputs['topk_indices'].int())
|