[Paddle Backend] Support GPU train/test/inference/LAMMPS with PaddlePaddle backend for water se_e2_a(revert code format) #4378

	on:
	push:
	pull_request:
	name: Test Python
	jobs:
	testpython:
	name: Test Python
	runs-on: ubuntu-latest
	strategy:
	matrix:
	include:
	- python: 3.7
	gcc: 5
	tf: 1.14
	- python: 3.7
	gcc: 6
	tf: 1.14
	- python: 3.7
	gcc: 7
	tf: 1.14
	- python: 3.7
	gcc: 8
	tf: 1.14
	- python: 3.8
	gcc: 8
	tf:
	- python: "3.11"
	gcc: 8
	tf:

	container: ghcr.io/deepmodeling/deepmd-kit-test-environment:py${{ matrix.python }}-gcc${{ matrix.gcc }}-tf${{ matrix.tf }}
	steps:
	- name: work around permission issue
	run: git config --global --add safe.directory /__w/deepmd-kit/deepmd-kit
	- uses: actions/checkout@v3
	# https://github.com/pypa/pip/issues/11770
	- run: python -m pip install -U "pip>=21.3.1,!=23.0.0"
	- run: pip install -e .[cpu,test]
	env:
	CC: gcc-${{ matrix.gcc }}
	CXX: g++-${{ matrix.gcc }}
	TENSORFLOW_VERSION: ${{ matrix.tf }}
	DP_BUILD_TESTING: 1
	- run: dp --version
	- run: pytest --cov=deepmd source/tests --durations=0
	- uses: codecov/codecov-action@v3
	with:
	gcov: true
	pass:
	name: Pass testing Python
	needs: [testpython]
	runs-on: ubuntu-latest
	if: always()
	steps:
	- name: Decide whether the needed jobs succeeded or failed
	uses: re-actors/alls-green@release/v1
	with:
	jobs: ${{ toJSON(needs) }}

Provide feedback