mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2026-04-23 00:17:25 +08:00
CSWYF3634076
9286403570
* support v1 loader * remove useless code * remove useless * [Model] support Qwen3VL images success * [Model] support Qwen3VL rope_3d * [Model] support Qwen3VL remove log * [Model] support Qwen3VL RL * [Model] support Qwen3VL tp * [Model] support Qwen3VL video * [Model] support Qwen3VL fix ernievl * [Model] support Qwen3VL fix get_image_boundaries.cc array out of bounds * [Model] support Qwen3VL fix multi card * [Model] support Qwen3VL file close * [Model] support Qwen3VL fix ce * [Model] support Qwen3VL fix unittest * [Model] support Qwen3VL add unittest --------- Co-authored-by: Ayakouji <yuhongh@qq.com>
78 lines
2.6 KiB
C++
78 lines
2.6 KiB
C++
// Copyright (c) 2025 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.
|
|
|
|
#include "paddle/extension.h"
|
|
|
|
#ifndef PD_BUILD_STATIC_OP
|
|
#define PD_BUILD_STATIC_OP(name) PD_BUILD_OP(static_op_##name)
|
|
#endif
|
|
|
|
std::vector<paddle::Tensor> GetImgBoundaries(
|
|
const paddle::Tensor& task_input_ids,
|
|
const paddle::Tensor& mm_num_token,
|
|
const int64_t image_patch_id) {
|
|
// All tensor in cpu
|
|
auto input_ids_ptr = task_input_ids.data<int64_t>();
|
|
int64_t seq_lens_origin = task_input_ids.numel();
|
|
auto mm_num_token_ptr = mm_num_token.data<int64_t>();
|
|
int64_t max_image_idx = mm_num_token.numel();
|
|
|
|
int token_idx = 0;
|
|
int image_idx = 0;
|
|
std::vector<int> img_boundaries, img_nums;
|
|
img_boundaries.emplace_back(0);
|
|
img_nums.emplace_back(0);
|
|
while (token_idx < seq_lens_origin) {
|
|
if (input_ids_ptr[token_idx] != image_patch_id) {
|
|
do {
|
|
token_idx++;
|
|
} while (token_idx < seq_lens_origin &&
|
|
input_ids_ptr[token_idx] != image_patch_id);
|
|
} else {
|
|
if (image_idx >= max_image_idx) {
|
|
token_idx = seq_lens_origin;
|
|
img_boundaries.emplace_back(token_idx);
|
|
img_nums.emplace_back(image_idx);
|
|
break;
|
|
}
|
|
int64_t cur_image_token_len = mm_num_token_ptr[image_idx];
|
|
image_idx++;
|
|
if (token_idx + cur_image_token_len > seq_lens_origin) {
|
|
token_idx = seq_lens_origin;
|
|
} else {
|
|
token_idx += cur_image_token_len;
|
|
}
|
|
}
|
|
img_boundaries.emplace_back(token_idx);
|
|
img_nums.emplace_back(image_idx);
|
|
}
|
|
|
|
int64_t num_img_boundaries = static_cast<int64_t>(img_boundaries.size());
|
|
auto out = paddle::full(
|
|
{2, num_img_boundaries}, 0, paddle::DataType::INT64, paddle::CPUPlace());
|
|
|
|
for (int i = 0; i < num_img_boundaries; i++) {
|
|
out.data<int64_t>()[i] = img_boundaries[i];
|
|
out.data<int64_t>()[num_img_boundaries + i] = img_nums[i];
|
|
}
|
|
|
|
return {out};
|
|
}
|
|
|
|
PD_BUILD_STATIC_OP(get_img_boundaries)
|
|
.Inputs({"task_input_ids", "mm_num_token"})
|
|
.Attrs({"image_patch_id: int64_t"})
|
|
.Outputs({"img_boundaries"})
|
|
.SetKernelFn(PD_KERNEL(GetImgBoundaries));
|