微调1.5之前的那个版本的模型

[rabbit12138-stack]

效仿孟加拉语微调步骤 https://github.com/PaddlePaddle/ERNIE/blob/release/v1.4/docs/paddleocr_vl_sft_zh.md
其中7.1推理环境配置在宿主机中安装 paddleocr 应该安装哪个版本的？
教程中给出的命令是 使用 python -m pip install -U "paddleocr[doc-parser]"和安装指定版本的 python -m pip install -U "paddleocr[doc-parser]"==3.3.0
均在验证模型的时候出现了权重方向 和当前 PaddleOCR-VL 推理加载器期望的方向不一致的问题。
以下是执行的错误信息
(.venv_paddleocr_v1) test@DESKTOP-DGS7RL:~$ paddleocr doc_parser -i testImage.png --vl_rec_model_name "PaddleOCR-VL-0.9B" --vl_rec_model_dir "./PaddleOCR-VL-SFT-Bengali" --save_path="./PaddleOCR-VL-SFT-Bengali_response"
Checking connectivity to the model hosters, this may take a while. To bypass this check, set DISABLE_MODEL_SOURCE_CHECK to True.
/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddle/utils/cpp_extension/extension_utils.py:718: UserWarning: No ccache found. Please be aware that recompiling all source files may be required. You can download and install ccache from: https://github.com/ccache/ccache/blob/master/doc/INSTALL.md
warnings.warn(warning_message)
Creating model: ('PP-DocLayoutV2', None)
Model files already exist. Using cached files. To redownload, please delete the directory manually: /home/test/.paddlex/official_models/PP-DocLayoutV2.
Creating model: ('PaddleOCR-VL-0.9B', './PaddleOCR-VL-SFT-Bengali')
Loading configuration file PaddleOCR-VL-SFT-Bengali/config.json
Loading weights file PaddleOCR-VL-SFT-Bengali/model.safetensors.index.json
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
use GQA - num_heads: 16- num_key_value_heads: 2
/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddle/utils/decorator_utils.py:420: Warning:
Non compatible API. Please refer to https://www.paddlepaddle.org.cn/documentation/docs/en/develop/guides/model_convert/convert_from_pytorch/api_difference/torch/torch.split.html first.
warnings.warn(
Traceback (most recent call last):
File "/home/test/.venv_paddleocr_v1/bin/paddleocr", line 8, in
sys.exit(console_entry())
^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddleocr/main.py", line 26, in console_entry
main()
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddleocr/_cli.py", line 192, in main
_execute(args)
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddleocr/_cli.py", line 181, in _execute
args.executor(args)
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddleocr/_pipelines/paddleocr_vl.py", line 363, in execute_with_args
perform_simple_inference(
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddleocr/_utils/cli.py", line 62, in perform_simple_inference
wrapper = wrapper_cls(**init_params)
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddleocr/_pipelines/paddleocr_vl.py", line 63, in init
super().init(**kwargs)
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddleocr/_pipelines/base.py", line 67, in init
self.paddlex_pipeline = self._create_paddlex_pipeline()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddleocr/_pipelines/base.py", line 105, in _create_paddlex_pipeline
return create_pipeline(config=self._merged_paddlex_config, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/pipelines/init.py", line 167, in create_pipeline
pipeline = BasePipeline.get(pipeline_name)(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/utils/deps.py", line 206, in _wrapper
return old_init_func(self, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/pipelines/_parallel.py", line 103, in init
self._pipeline = self._create_internal_pipeline(config, self.device)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/pipelines/_parallel.py", line 158, in _create_internal_pipeline
return self._pipeline_cls(
^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/pipelines/paddleocr_vl/pipeline.py", line 130, in init
self.vl_rec_model = self.create_model(vl_rec_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/pipelines/base.py", line 106, in create_model
model = create_predictor(
^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/models/init.py", line 84, in create_predictor
return BasePredictor.get(model_name)(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/models/doc_vlm/predictor.py", line 62, in init
self.infer, self.processor = self._build(**kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/models/doc_vlm/predictor.py", line 148, in _build
model = PaddleOCRVLForConditionalGeneration.from_pretrained(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/models/common/vlm/transformers/model_utils.py", line 1902, in from_pretrained
cls._load_pretrained_model(
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/models/common/vlm/transformers/model_utils.py", line 1592, in _load_pretrained_model
error_msgs += _load_state_dict_into_model(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/models/common/vlm/transformers/model_utils.py", line 299, in _load_state_dict_into_model
model_to_load.set_hf_state_dict(state_dict)
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/models/doc_vlm/modeling/paddleocr_vl/_paddleocr_vl.py", line 618, in set_hf_state_dict
state_dict = _convert_state_dict(state_dict)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddlex/inference/models/doc_vlm/modeling/paddleocr_vl/_paddleocr_vl.py", line 613, in _convert_state_dict
paddle.concat([q_proj, k_proj, v_proj], axis=-1)
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddle/utils/decorator_utils.py", line 50, in wrapper
return func(*processed_args, **processed_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/test/.venv_paddleocr_v1/lib/python3.12/site-packages/paddle/tensor/manipulation.py", line 1522, in concat
return _C_ops.concat(input, axis, out=out)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: (InvalidArgument) The 0-th dimension of input[0] and input[1] is expected to be equal.But received input[0]'s shape = [2048, 1024], input[1]'s shape = [256, 1024].
[Hint: Expected inputs_dims[0][j] == inputs_dims[i][j], but received inputs_dims[0][j]:2048 != inputs_dims[i][j]:256.] (at /paddle/paddle/phi/kernels/funcs/concat_funcs.h:72)

(.venv_paddleocr_v1) test@DESKTOP-DGS7RL:~$

安装的PaddlePaddle 的版本是
python -m pip install paddlepaddle-gpu==3.2.0 -i https://www.paddlepaddle.org.cn/packages/stable/cu126/
python的版本是3.12.3
CUDA为12.6

[2742195759]

您好，我是熊昆，您的邮件我已收到，我会尽快回复谢谢。

[rabbit12138-stack]

您好，我是熊昆，您的邮件我已收到，我会尽快回复谢谢。

补充用到的相关依赖是
python -m pip list | grep -E "paddleocr|paddlex|paddlepaddle|safetensors"
paddleocr 3.3.2
paddlepaddle-gpu 3.2.1
paddlex 3.3.10
safetensors 0.6.2.dev0