【PaddleNLP No.6】Fix text_classification PIR

[hanlintang]

PR types

Bug fixes

PR changes

Models

Description

1. 添加text_classification中rretrieval_based的hierarchical/multi_class/multi_label模型基于paddle.inference推理对PIR的适配；
2. 修改设计到导出模型的文档。
   修改部分可以成功运行推理。

I0425 05:49:54.312034 17340 pir_interpreter.cc:1640] pir interpreter is running by trace mode ...
(1, 768)
......
   5.13511896e-02  1.27085205e-02 -1.99236199e-02 -3.18305828e-02
  -6.04303088e-03 -2.46610641e-02 -1.73023611e-03 -6.33593462e-03
  -1.72202699e-02  1.04825031e-02 -5.76227345e-03 -2.10440122e-02
   4.56353836e-02 -1.11948913e-02  3.09750624e-02  4.88217883e-02
   6.00924827e-02 -1.12980399e-02  2.02461407e-02  4.56624702e-02
  -2.55786590e-02  3.89200114e-02 -9.09841619e-03  4.24942374e-02
  -6.72577918e-02  7.34342411e-02  2.94713527e-02 -5.68274520e-02
   5.37596755e-02 -2.73752045e-02  2.10331194e-02  4.71913368e-02]]
[0.6664873361587524, 0.4887930154800415]

Modified:

1. slm/applications/text_classification/hierarchical/retrieval_based/deploy/python/predict.py
2. slm/applications/text_classification/hierarchical/retrieval_based/utils/feature_extract.py
3. slm/applications/text_classification/multi_class/retrieval_based/deploy/python/predict.py
4. slm/applications/text_classification/multi_class/retrieval_based/utils/feature_extract.py
5. slm/applications/text_classification/multi_label/README.md
6. slm/applications/text_classification/multi_label/retrieval_based/deploy/python/predict.py
7. slm/applications/text_classification/multi_label/retrieval_based/utils/feature_extract.py

其他问题

1. 本次修改验证了三个模型的训练、导出及paddle.inference推理。没有修改paddle_serving, simple_serving, triton_serving及onnx部署的内容，因为其已经无法与paddle3.0.0适配，如onnx部署：

[2025-04-24 11:21:34,047] [    INFO] - >>> [InferBackend] Creating Engine ...
Traceback (most recent call last):
  File "infer.py", line 69, in <module>
    predictor = Predictor(args, label_list)
  File "/home/aistudio/PaddleNLP/slm/applications/text_classification/hierarchical/deploy/predictor/predictor.py", line 121, in __init__
    self.inference_backend = InferBackend(
  File "/home/aistudio/PaddleNLP/slm/applications/text_classification/hierarchical/deploy/predictor/predictor.py", line 35, in __init__
    onnx_model = paddle2onnx.export(
TypeError: export() got an unexpected keyword argument 'model_file'

2. 没有修改few-shot的部分，根据文档该部分将在未来删除，并推荐使用其他zero-shot模型。

Issue: #9763
@DrownFish19

[paddle-bot]

Thanks for your contribution!

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 48.67%. Comparing base (323ffec) to head (f8d612b).
Report is 27 commits behind head on develop.

❌ Your project status has failed because the head coverage (48.67%) is below the target coverage (58.00%). You can increase the head coverage or adjust the target coverage.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop   #10497      +/-   ##
===========================================
- Coverage    48.68%   48.67%   -0.01%     
===========================================
  Files          768      768              
  Lines       126915   126915              
===========================================
- Hits         61785    61777       -8     
- Misses       65130    65138       +8     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[DrownFish19]

LGTM