neo4j-graphrag 1.8.0 requires numpy<3.0.0,>=2.0.0

[carnak]

Am finding it difficult / impossible to include neo4j-graphrag in our jupyterlab k8s pod, as it really causes conflicts w/ pandas, numpy, etc. Here is the system response to doing 'python -m pip install --upgrade pip':

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
neo4j-graphrag 1.8.0 requires numpy<3.0.0,>=2.0.0; python_version >= "3.9" and python_version < "3.13", but you have numpy 1.26.4 which is incompatible.
numba 0.57.1 requires numpy<1.25,>=1.21, but you have numpy 1.26.4 which is incompatible.

[stellasia]

Hi carnak,

The version of numba creating conflicts seems to be quite old (released two years ago?). Can you try to upgrade it before we investigate this further?

Thanks

[sivajik34]

neo4j-graphrag 1.8.0 requires NumPy ≥ 2.0.0

spaCy / thinc are not compatible with NumPy 2.x

[stellasia]

Hi @sivajik34 ,

Can you help me understand this better, because what I see in the dependencies of these packages is (last available tag):

• for spacy, numpy>=1.19.0; python_version >= "3.9"
• for thinc: numpy>=1.19.0

So no condition on numpy < 2.0?