paddlepaddle3.3.1 + paddledetectionv2.9.0 导出模型失败

[tangchang1988]

问题确认 Search before asking

• 我已经查询历史issue，没有发现相似的bug。I have searched the issues and found no similar bug report.

Bug组件 Bug Component

No response

Bug描述 Describe the Bug

训练完成后进行模型导出操作，没有对源码做任何改动，错误日志如下：

root@b9ab843343e9:/paddle/workspace/PaddleDetection# python tools/export_model.py -c configs/my_ppyoloe_plus.yml
-o weights=output/ppyoloe_plus/best_model.pdparams
TestReader.inputs_def.image_shape=[3,640,640]
--output_dir=./inference_model
[07/31 05:18:40] ppdet.utils.checkpoint INFO: Skipping import of the encryption module.
Warning: import ppdet from source directory without installing, run 'python setup.py install' to install ppdet firstly
[07/31 05:18:41] ppdet.utils.checkpoint INFO: Finish loading model weights: output/ppyoloe_plus/best_model.pdparams
W0731 05:18:42.064301 72834 eager_utils.cc:3584] Paddle static graph(PIR) not support input out tensor for now!!!!!
loading annotations into memory...
Done (t=0.00s)
creating index...
index created!
[07/31 05:18:43] ppdet.engine INFO: Export inference config file to ./inference_model/my_ppyoloe_plus/infer_cfg.yml
[07/31 05:18:43] ppdet.engine INFO: Skipping import of the encryption module.
Traceback (most recent call last):
File "/paddle/workspace/PaddleDetection/tools/export_model.py", line 121, in
main()
File "/paddle/workspace/PaddleDetection/tools/export_model.py", line 118, in main
run(FLAGS, cfg)
File "/paddle/workspace/PaddleDetection/tools/export_model.py", line 80, in run
trainer.export(FLAGS.output_dir, for_fd=FLAGS.for_fd)
File "/paddle/workspace/PaddleDetection/ppdet/engine/trainer.py", line 1458, in export
static_model, pruned_input_spec = self._model_to_static(
File "/paddle/workspace/PaddleDetection/ppdet/engine/trainer.py", line 1299, in _model_to_static
input_spec, static_model.forward.main_program,
File "/usr/local/lib/python3.10/dist-packages/paddle/jit/dy2static/program_translator.py", line 1101, in main_program
concrete_program = self.concrete_program
File "/usr/local/lib/python3.10/dist-packages/paddle/jit/dy2static/program_translator.py", line 985, in concrete_program
return self.concrete_program_specify_input_spec(input_spec=None)
File "/usr/local/lib/python3.10/dist-packages/paddle/jit/dy2static/program_translator.py", line 1029, in concrete_program_specify_input_spec
concrete_program, _ = self.get_concrete_program(
File "/usr/local/lib/python3.10/dist-packages/paddle/jit/dy2static/program_translator.py", line 918, in get_concrete_program
concrete_program, partial_program_layer = self._program_cache[
File "/usr/local/lib/python3.10/dist-packages/paddle/jit/dy2static/program_translator.py", line 1690, in getitem
self._caches[item_id] = self._build_once(item)
File "/usr/local/lib/python3.10/dist-packages/paddle/jit/dy2static/program_translator.py", line 1635, in _build_once
concrete_program = ConcreteProgram.from_func_spec(
File "/usr/local/lib/python3.10/dist-packages/paddle/base/wrapped_decorator.py", line 50, in wrapper
return decorated(*args, **kwargs)
File "/usr/local/lib/python3.10/dist-packages/paddle/base/dygraph/base.py", line 114, in impl
return func(*args, **kwargs)
File "/usr/local/lib/python3.10/dist-packages/paddle/jit/dy2static/program_translator.py", line 1407, in from_func_spec
error_data.raise_new_exception()
File "/usr/local/lib/python3.10/dist-packages/paddle/jit/dy2static/error.py", line 454, in raise_new_exception
raise new_exception from None
TypeError: In transformed code:

File "/paddle/workspace/PaddleDetection/ppdet/modeling/architectures/meta_arch.py", line 55, in forward
    if self.training:
File "/paddle/workspace/PaddleDetection/ppdet/modeling/architectures/meta_arch.py", line 65, in forward
    for inp in inputs_list:
File "/paddle/workspace/PaddleDetection/ppdet/modeling/architectures/meta_arch.py", line 69, in forward
    outs.append(self.get_pred())
File "/paddle/workspace/PaddleDetection/ppdet/modeling/architectures/yolo.py", line 150, in get_pred
    return self._forward()
File "/paddle/workspace/PaddleDetection/ppdet/modeling/architectures/yolo.py", line 92, in _forward
    if self.training:
File "/paddle/workspace/PaddleDetection/ppdet/modeling/architectures/yolo.py", line 101, in _forward
    yolo_head_outs = self.yolo_head(neck_feats)
File "/paddle/workspace/PaddleDetection/ppdet/modeling/heads/ppyoloe_head.py", line 277, in forward
    if self.training:
File "/paddle/workspace/PaddleDetection/ppdet/modeling/heads/ppyoloe_head.py", line 288, in forward
    return self.forward_eval(feats)
File "/paddle/workspace/PaddleDetection/ppdet/modeling/heads/ppyoloe_head.py", line 242, in forward_eval
    for i, feat in enumerate(feats):
File "/paddle/workspace/PaddleDetection/ppdet/modeling/heads/ppyoloe_head.py", line 249, in forward_eval
    cls_logit = self.pred_cls[i](self.stem_cls[i](feat, avg_feat) +
File "/paddle/workspace/PaddleDetection/ppdet/modeling/heads/ppyoloe_head.py", line 50, in forward
    def forward(self, feat, avg_feat):
        weight = F.sigmoid(self.fc(avg_feat))
        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ <--- HERE
        if self.conv:
            return self.conv(feat * weight)

File "/usr/local/lib/python3.10/dist-packages/paddle/base/dygraph/generated_tensor_methods_patch.py", line 106, in _sigmoid
    return _C_ops.sigmoid(*args, **kwargs)

TypeError: (InvalidType) sigmoid(): argument (position 1) must be Value, but got Variable (at /paddle/paddle/fluid/pybind/eager_utils.cc:2308)

复现环境 Environment

• OS: ubuntn 24.0.4
• PaddlePaddle: 3.3.1
• PaddleDetection: release/2.9
• Python: 3.10.12
• CUDA: 13.0
• CUDNN: 9.13.0
• GCC: 13.1.0

Bug描述确认 Bug description confirmation

• 我确认已经提供了Bug复现步骤、代码改动说明、以及环境信息，确认问题是可以复现的。I confirm that the bug replication steps, code change instructions, and environment information have been provided, and the problem can be reproduced.

是否愿意提交PR？ Are you willing to submit a PR?

• 我愿意提交PR！I'd like to help by submitting a PR!