Unpinned numpy breaks vectorcode on Mac OS

[kwrobert]

When following the installation instructions for the vectorcode CLI on my Mac OS environment, vectorcode does not work. If I downgrade numpy to numpy==1.26.4 it fixes the issue.

Details of my environment:

OS: Mac OS 14.7.3 (23H417)
Python: 3.12.7
pipx version: 1.7.1
pip version: 24.1.2

Here is the error I get with the default installation:

$ vectorcode vectorise ./lua/codecompanion/init.lua
Host at 127.0.0.1:8000 is unavailable. VectorCode will start its own Chromadb at a random port.
Creating database at /Users/kyle/.local/share/vectorcode/chromadb/.

A module that was compiled using NumPy 1.x cannot be run in
NumPy 2.2.3 as it may crash. To support both 1.x and 2.x
versions of NumPy, modules must be compiled with NumPy 2.0.
Some module may need to rebuild instead e.g. with 'pybind11>=2.12'.

If you are a user of the module, the easiest solution will be to
downgrade to 'numpy<2' or try to upgrade the affected module.
We expect that some modules will need time to support NumPy 2.

Traceback (most recent call last):  File "/Users/kyle/.local/bin/vectorcode", line 8, in <module>
    sys.exit(main())
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/vectorcode/main.py", line 90, in main
    return asyncio.run(async_main())
  File "/Users/kyle/.local/share/mise/installs/python/3.12.9/lib/python3.12/asyncio/runners.py", line 195, in run
    return runner.run(main)
  File "/Users/kyle/.local/share/mise/installs/python/3.12.9/lib/python3.12/asyncio/runners.py", line 118, in run
    return self._loop.run_until_complete(task)
  File "/Users/kyle/.local/share/mise/installs/python/3.12.9/lib/python3.12/asyncio/base_events.py", line 678, in run_until_complete
    self.run_forever()
  File "/Users/kyle/.local/share/mise/installs/python/3.12.9/lib/python3.12/asyncio/base_events.py", line 645, in run_forever
    self._run_once()
  File "/Users/kyle/.local/share/mise/installs/python/3.12.9/lib/python3.12/asyncio/base_events.py", line 1999, in _run_once
    handle._run()
  File "/Users/kyle/.local/share/mise/installs/python/3.12.9/lib/python3.12/asyncio/events.py", line 88, in _run
    self._context.run(self._callback, *self._args)
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/vectorcode/main.py", line 70, in async_main
    return_val = await vectorise(final_configs)
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/vectorcode/subcommands/vectorise.py", line 87, in vectorise
    collection = await make_or_get_collection(client, configs)
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/vectorcode/common.py", line 141, in make_or_get_collection
    embedding_function=get_embedding_function(configs),
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/vectorcode/common.py", line 116, in get_embedding_function
    return getattr(embedding_functions, configs.embedding_function)(
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/utils/embedding_functions/sentence_transformer_embedding_function.py", line 32, in __init__
    from sentence_transformers import SentenceTransformer
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/sentence_transformers/__init__.py", line 9, in <module>
    from sentence_transformers.backend import (
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/sentence_transformers/backend.py", line 11, in <module>
    from sentence_transformers.util import disable_datasets_caching, is_datasets_available
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/sentence_transformers/util.py", line 17, in <module>
    import torch
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/torch/__init__.py", line 1477, in <module>
    from .functional import *  # noqa: F403
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/torch/functional.py", line 9, in <module>
    import torch.nn.functional as F
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/torch/nn/__init__.py", line 1, in <module>
    from .modules import *  # noqa: F403
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/torch/nn/modules/__init__.py", line 35, in <module>
    from .transformer import TransformerEncoder, TransformerDecoder, \
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/torch/nn/modules/transformer.py", line 20, in <module>
    device: torch.device = torch.device(torch._C._get_default_device()),  # torch.device('cpu'),
/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/torch/nn/modules/transformer.py:20: UserWarning: Failed to initialize NumPy: _ARRAY_API not found (Triggered internally at /Users/runner/work/pytorch/pytorch/pytorch/torch/csrc/utils/tensor_numpy.cpp:84.)
  device: torch.device = torch.device(torch._C._get_default_device()),  # torch.device('cpu'),
Vectorising files...:   0%|                                                                                                                                                                               | 0/1 [00:01<?, ?it/s]
Traceback (most recent call last):
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 90, in wrapper
    return func(self, *args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 213, in _validate_and_prepare_add_request
    add_embeddings = self._embed_record_set(record_set=add_records)
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 526, in _embed_record_set
    return self._embed(input=record_set[field])  # type: ignore[literal-required]
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 539, in _embed
    return self._embedding_function(input=input)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/types.py", line 466, in __call__
    result = call(self, input)
             ^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/utils/embedding_functions/sentence_transformer_embedding_function.py", line 48, in __call__
    for embedding in self._model.encode(
                     ^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/sentence_transformers/SentenceTransformer.py", line 674, in encode
    all_embeddings = np.asarray([emb.numpy() for emb in all_embeddings])
                                 ^^^^^^^^^^^
RuntimeError: Numpy is not available

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/vectorcode/main.py", line 70, in async_main
    return_val = await vectorise(final_configs)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/vectorcode/subcommands/vectorise.py", line 130, in vectorise
    await task
  File "/Users/kyle/.local/share/mise/installs/python/3.12.9/lib/python3.12/asyncio/tasks.py", line 631, in _wait_for_one
    return f.result()  # May raise f.exception().
           ^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/vectorcode/subcommands/vectorise.py", line 62, in chunked_add
    await collection.add(
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/models/AsyncCollection.py", line 63, in add
    add_request = self._validate_and_prepare_add_request(
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 93, in wrapper
    raise type(e)(msg).with_traceback(e.__traceback__)
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 90, in wrapper
    return func(self, *args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 213, in _validate_and_prepare_add_request
    add_embeddings = self._embed_record_set(record_set=add_records)
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 526, in _embed_record_set
    return self._embed(input=record_set[field])  # type: ignore[literal-required]
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 539, in _embed
    return self._embedding_function(input=input)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/api/types.py", line 466, in __call__
    result = call(self, input)
             ^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/chromadb/utils/embedding_functions/sentence_transformer_embedding_function.py", line 48, in __call__
    for embedding in self._model.encode(
                     ^^^^^^^^^^^^^^^^^^^
  File "/Users/kyle/.local/pipx/venvs/vectorcode/lib/python3.12/site-packages/sentence_transformers/SentenceTransformer.py", line 674, in encode
    all_embeddings = np.asarray([emb.numpy() for emb in all_embeddings])
                                 ^^^^^^^^^^^
RuntimeError: Numpy is not available in add.
vectorcode vectorise ./lua/codecompanion/init.lua  8.15s user 2.30s system 14% cpu 1:10.72 total

If I downgrade numpy in the pipx virtualenv via: pipx runpip vectorcode install numpy==1.26.4

Then everything works:

ectorcode vectorise ./lua/codecompanion/init.lua
Host at 127.0.0.1:8000 is unavailable. VectorCode will start its own Chromadb at a random port.
Creating database at /Users/kyle/.local/share/vectorcode/chromadb/.
Vectorising files...: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:01<00:00,  1.68s/it]
  Added    Updated    Removed
-------  ---------  ---------
      1          0          0
vectorcode vectorise ./lua/codecompanion/init.lua  9.85s user 3.18s system 56% cpu 23.140 total

I suspect pinning numpy in the pyproject.toml file for this repo will fix the issue.

[Davidyz]

Hi, thanks for raising this issue! I could not reproduce this on Python 3.12.9 on linux with numpy==2.1.3, but since the error was triggered by sentence transformer and torch, could you check their versions too? There's a chance that you're using an old version of either of these 2, and they may not compatible with numpy 2.0+. I prefer to stick to newer releases where possible, so if upgrading to newer torch/sentence transformer solves the issue, I prefer this solution instead.

[kwrobert]

Interesting! What operating system are you running on? Linux? And what CPU architecture? I'm on an Intel Mac at the moment. It's possible for Python package versions to only be available on certain operating system+CPU architecture combinations.

Here are the versions you asked for:

$ pipx runpip vectorcode freeze | grep torch
torch==2.2.2

$ pipx runpip vectorcode freeze | grep sentence
sentence-transformers==3.4.1

Here is the full version loadout in the vectorcode virtual environment created by pipx:

$ pipx runpip vectorcode freeze
aiohappyeyeballs==2.4.6
aiohttp==3.11.12
aiosignal==1.3.2
annotated-types==0.7.0
anyio==4.8.0
asgiref==3.8.1
attrs==25.1.0
backoff==2.2.1
bcrypt==4.2.1
build==1.2.2.post1
cachetools==5.5.1
certifi==2025.1.31
charset-normalizer==3.4.1
chroma-hnswlib==0.7.6
chromadb==0.6.3
click==8.1.8
coloredlogs==15.0.1
Deprecated==1.2.18
durationpy==0.9
fastapi==0.115.8
filelock==3.17.0
flatbuffers==25.2.10
frozenlist==1.5.0
fsspec==2025.2.0
google-auth==2.38.0
googleapis-common-protos==1.67.0
grpcio==1.70.0
h11==0.14.0
httpcore==1.0.7
httptools==0.6.4
httpx==0.28.1
huggingface-hub==0.28.1
humanfriendly==10.0
idna==3.10
importlib_metadata==8.5.0
importlib_resources==6.5.2
Jinja2==3.1.5
joblib==1.4.2
kubernetes==32.0.0
markdown-it-py==3.0.0
MarkupSafe==3.0.2
mdurl==0.1.2
mmh3==5.1.0
monotonic==1.6
mpmath==1.3.0
multidict==6.1.0
networkx==3.4.2
numpy==1.26.4
oauthlib==3.2.2
onnxruntime==1.20.1
opentelemetry-api==1.30.0
opentelemetry-exporter-otlp-proto-common==1.30.0
opentelemetry-exporter-otlp-proto-grpc==1.30.0
opentelemetry-instrumentation==0.51b0
opentelemetry-instrumentation-asgi==0.51b0
opentelemetry-instrumentation-fastapi==0.51b0
opentelemetry-proto==1.30.0
opentelemetry-sdk==1.30.0
opentelemetry-semantic-conventions==0.51b0
opentelemetry-util-http==0.51b0
orjson==3.10.15
overrides==7.7.0
packaging==24.2
pathspec==0.12.1
pillow==11.1.0
posthog==3.13.0
propcache==0.2.1
protobuf==5.29.3
psutil==7.0.0
pyasn1==0.6.1
pyasn1_modules==0.4.1
pydantic==2.10.6
pydantic_core==2.27.2
Pygments==2.19.1
PyPika==0.48.9
pyproject_hooks==1.2.0
python-dateutil==2.9.0.post0
python-dotenv==1.0.1
PyYAML==6.0.2
regex==2024.11.6
requests==2.32.3
requests-oauthlib==2.0.0
rich==13.9.4
rsa==4.9
safetensors==0.5.2
scikit-learn==1.6.1
scipy==1.15.1
sentence-transformers==3.4.1
setuptools==75.8.0
shellingham==1.5.4
shtab==1.7.1
six==1.17.0
sniffio==1.3.1
starlette==0.45.3
sympy==1.13.3
tabulate==0.9.0
tenacity==9.0.0
threadpoolctl==3.5.0
tokenizers==0.21.0
torch==2.2.2
tqdm==4.67.1
transformers==4.48.3
typer==0.15.1
typing_extensions==4.12.2
urllib3==2.3.0
uvicorn==0.34.0
uvloop==0.21.0
VectorCode==0.3.0
watchfiles==1.0.4
websocket-client==1.8.0
websockets==15.0
wheel==0.45.1
wrapt==1.17.2
yarl==1.18.3
zipp==3.21.0

[Davidyz]

Hi, thanks for the updates. Can you try upgrading torch to 2.4.1 or later, and numpy to 2.1.1 or later? I found this issue and it appears that this has been resolved in later versions. If this solves your issue, I'll fix this issue by adding minimum version constraints for both numpy and torch.

[kwrobert]

I think the problem is that torch==2.4.1 is not available for my platform. When I attempt to upgrade torch here is what I get:

$ pipx runpip vectorcode install --upgrade torch==2.4.1
ERROR: Could not find a version that satisfies the requirement torch==2.4.1 (from versions: 2.2.0, 2.2.1, 2.2.2)
ERROR: No matching distribution found for torch==2.4.1

if you look at the list of available wheels for torch==2.4.1, you can see they don't pre-build wheels for Intel Macs anymore (which is fair, the Mac I'm running on is from 2019). So that is the issue here, I'm running on an Intel Mac and numpy and pytorch no longer publish pre-built wheels for that platform+architecture combo in their most recent versions.

[kwrobert]

So anyway, I think the most reasonable thing to do here is just document somewhere that Intel Macs will need to pin to older versions of torch and/or numpy and move on. Everything seems to work fine when doing that, and Intel Macs are a edge case, slightly outdated platform that probably isn't worth adding extra configuration to your build system to support.

[Davidyz]

Yes, I'm testing the viability of adding an extra dependency group legacy with numpy version constrained to <2.0.0 so that people on an older system can just do pipx install vectorcode[legacy], which avoids manually pinning the versions. Sorry for the inconvenience.

[kwrobert]

No problem, sounds like a reasonable approach. Pip requirement specifiers might also be useful to you here. You can use those to only install a package which is pinned to a particular version for specific platforms.

[Davidyz]

I had some unsuccessful experiences using these with pdm for a different dependency issue. I'll keep this in mind, but at the moment I'll close this because pinning numpy and torch versions works for now.