diff --git a/science/py-kliff/Makefile b/science/py-kliff/Makefile
index 52f3f8368887..023e89202e43 100644
--- a/science/py-kliff/Makefile
+++ b/science/py-kliff/Makefile
@@ -1,38 +1,54 @@
 PORTNAME=	kliff
-DISTVERSION=	0.4.4
-PORTREVISION=	3
+DISTVERSIONPREFIX=	v
+DISTVERSION=	1.0.1-45
+DISTVERSIONSUFFIX=	-gdf2d4de
 CATEGORIES=	science python # chemistry
-MASTER_SITES=	PYPI
+#MASTER_SITES=	PYPI # release 1.0.1 is faulty: it has the numpy<2 requirement
 PKGNAMEPREFIX=	${PYTHON_PKGNAMEPREFIX}
 
 MAINTAINER=	yuri@FreeBSD.org
 COMMENT=	KIM-based Learning-Integrated Fitting Framework
 WWW=		https://kliff.readthedocs.io/en/latest/ \
 		https://github.com/openkim/kliff
 
 LICENSE=	LGPL21
 LICENSE_FILE=	${WRKSRC}/LICENSE
 
 BUILD_DEPENDS=	${PYTHON_PKGNAMEPREFIX}pybind11>0:devel/py-pybind11@${PY_FLAVOR}
-RUN_DEPENDS=	${PYTHON_PKGNAMEPREFIX}loguru>0:devel/py-loguru@${PY_FLAVOR} \
+RUN_DEPENDS=	${PYTHON_PKGNAMEPREFIX}ase>0:science/py-ase@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}loguru>0:devel/py-loguru@${PY_FLAVOR} \
 		${PYTHON_PKGNAMEPREFIX}monty>0:devel/py-monty@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}numpy>0:math/py-numpy@${PY_FLAVOR} \
 		${PYTHON_PKGNAMEPREFIX}requests>0:www/py-requests@${PY_FLAVOR} \
 		${PYTHON_PKGNAMEPREFIX}scipy>0:science/py-scipy@${PY_FLAVOR} \
 		${PYTHON_PKGNAMEPREFIX}pyyaml>0:devel/py-pyyaml@${PY_FLAVOR}
-RUN_DEPENDS+=	${PYTHON_PKGNAMEPREFIX}emcee>0:math/py-emcee@${PY_FLAVOR} \
-		${PYTHON_PKGNAMEPREFIX}numpy>=1.16:math/py-numpy@${PY_FLAVOR} \
-		${PYTHON_PKGNAMEPREFIX}kimpy>0:science/py-kimpy@${PY_FLAVOR} \
+RUN_DEPENDS_torch=	\
 		${PYTHON_PKGNAMEPREFIX}pytorch>0:misc/py-pytorch@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}pytorch-lightning>0:misc/py-pytorch-lightning@${PY_FLAVOR}
+		# missing: torch_geometric, torch_scatter, tensorboard, tensorboardx
+RUN_DEPENDS+=	${RUN_DEPENDS_torch}
+TEST_DEPENDS+=	${PYTHON_PKGNAMEPREFIX}emcee>0:math/py-emcee@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}kimpy>0:science/py-kimpy@${PY_FLAVOR} \
+		${PYTHON_PKGNAMEPREFIX}pytest-cov>0:devel/py-pytest-cov@${PY_FLAVOR}
+		# libdescriptor is missing
+		# also need 'torch' dependency
 
 USES=		compiler:c++11-lang python
 USE_PYTHON=	distutils autoplist pytest # tests fail to run, see https://github.com/openkim/kliff/issues/81, and https://github.com/openkim/kliff/issues/197
 
+USE_GITHUB=	yes
+GH_ACCOUNT=	openkim
+
 TEST_ENV=	${MAKE_ENV} PYTHONPATH=${STAGEDIR}${PYTHONPREFIX_SITELIBDIR}
 TEST_WRKSRC=	${WRKSRC}/tests
+TEST_ARGS=	--disable-plugin-autoload
 
 post-install:
 	@${FIND} ${STAGEDIR}${PYTHONPREFIX_SITELIBDIR} -name "*.so" | ${XARGS} ${STRIP_CMD}
 
+test-quick:
+	@cd ${TEST_WRKSRC} && ${SETENV} ${TEST_ENV} ${PYTHON_CMD} ${FILESDIR}/example.py
+
 # tests as of 0.4.4: 17 failed, 29 passed, 16 errors in 14.58s, see https://github.com/openkim/kliff/issues/210
 
 .include <bsd.port.mk>
diff --git a/science/py-kliff/distinfo b/science/py-kliff/distinfo
index 4faca49b1df5..8c15976866dd 100644
--- a/science/py-kliff/distinfo
+++ b/science/py-kliff/distinfo
@@ -1,3 +1,3 @@
-TIMESTAMP = 1744348783
-SHA256 (kliff-0.4.4.tar.gz) = 0644eebadfe21b77389eaa6e001ebb166bd95ed06993c309293190991ff76980
-SIZE (kliff-0.4.4.tar.gz) = 3172741
+TIMESTAMP = 1790233395
+SHA256 (openkim-kliff-v1.0.1-45-gdf2d4de_GH0.tar.gz) = c32dbff4d1118dfa8f64ecb2f89520047d405e91a270f3a2b6e0c6f1480f80f6
+SIZE (openkim-kliff-v1.0.1-45-gdf2d4de_GH0.tar.gz) = 3846602
diff --git a/science/py-kliff/files/example.py b/science/py-kliff/files/example.py
new file mode 100644
index 000000000000..f7595a28607a
--- /dev/null
+++ b/science/py-kliff/files/example.py
@@ -0,0 +1,44 @@
+from kliff.dataset import Dataset
+from kliff.models import NeuralNetwork
+from kliff.utils import download_dataset
+
+# 1. Using the legacy/imperative modules for custom configuration setup
+from kliff.legacy import nn
+from kliff.legacy.calculators import CalculatorTorch
+from kliff.legacy.descriptors import SymmetryFunction
+from kliff.legacy.loss import Loss
+
+# 2. Define a descriptor to featurize atomic environments (e.g., for Silicon)
+descriptor = SymmetryFunction(
+    cut_name="cos",
+    cut_dists={"Si-Si": 5.0},
+    hyperparams="set51",
+    normalize=True
+)
+
+# 3. Create a multi-layer Neural Network potential using the descriptor
+model = NeuralNetwork(descriptor)
+model.add_layers(
+    nn.Linear(descriptor.get_size(), 10),  # First hidden layer (10 units)
+    nn.Tanh(),
+    nn.Linear(10, 10),                     # Second hidden layer (10 units)
+    nn.Tanh(),
+    nn.Linear(10, 1),                      # Output layer (Energy prediction)
+)
+
+# 4. Download and parse a crystal structure training dataset
+dataset_path = download_dataset(dataset_name="Si_training_set")
+dataset_path = dataset_path.joinpath("varying_alat")
+train_set = Dataset.from_path(dataset_path)
+configs = train_set.get_configs()
+
+# 5. Initialize the calculator (using PyTorch backend) to compute energy/forces
+calc = CalculatorTorch(model, gpu=False)
+calc.create(configs)
+
+# 6. Define the optimizer loss function
+loss = Loss(calc)
+
+print(f"Total configurations loaded: {len(configs)}")
+print("KLIFF environment successfully configured for potential fitting.")
+