Open Spadet opened 2 years ago
Hello Spadet, i'm also still struggling with dependency conflicts mostly from click, requests, spacy, urllib3, and others. So a solution to your problem would be super helpful. A colleague of mine also tried the installation of 2.1.1 and 2.1.2 without success on ubuntu and windows.
greetings viktor
Good to know I'm not the only one facing problems with the latest version! I managed to install 2.1.2 on Ubuntu 20.04 using a Python 3.7.11 environment. I also installed jsonnet beforehand and installed GCC at one point too. Hope it will help!
Thanks for the info, i also have exactly one cde 2.1.1 installation in a venv environment on ubuntu 20.04. But i couldnt replicate the installation in conda for example or another venv environment.
Sorry for the late reply, does this seem to happen when CDE is installed in a clean environment or only if you already have other things installed? There's a difficulty in that the NER code uses allennlp 0.9.0 so we can't use newer versions of some libraries, probably leading to these conflicts... If the issue happens in a clean environment, I'm happy to look into it, otherwise I don't think there's a solution in the near term... If anyone wants to make a PR migrating the code and the NER in particular to use more recent versions of allennlp it would be much appreciated; I would love to do this but currently don't have the time
Indeed, I had a clean environment when trying the two notebooks so it was surprising!
I'll try installing on a Colab notebook then - my suspicion is that it comes with some libraries pre-installed for the average usecase, but those are newer versions than what CDE wants due to its dependence on allennlp 0.9.0
This can be the reason you're right! However, I did encounter similar installation issues using a clean Python 3.9.X environment although it is said to be compatible (OS: Ubuntu 20.04).
Yeah there seems to be issues with Python 3.9.x, I'll have to remove it from the readme saying it's compatible...
What environment is known to work? After trying in colab and recieving the above dependency lock i tried in a virtual environment with python 3.6 running into issues building as the tokenizers package seems to rely on pyo3 which requires at least 3.7:
error: failed to run custom build command for
pyo3-ffi v0.16.6
Caused by: process didn't exit successfully:
/tmp/pip-install-vnmpfxfz/tokenizers_32f224e543a24d05a5fefeb3c8a79c1b/target/release/build/pyo3-ffi-31b85eb77d016a47/build-script-build
(exit status: 1) --- stdout cargo:rerun-if-env-changed=PYO3_CROSS cargo:rerun-if-env-changed=PYO3_CROSS_LIB_DIR cargo:rerun-if-env-changed=PYO3_CROSS_PYTHON_VERSION cargo:rerun-if-env-changed=PYO3_CROSS_PYTHON_IMPLEMENTATION cargo:rerun-if-env-changed=PYO3_PRINT_CONFIG--- stderr error: the configured Python interpreter version (3.6) is lower than PyO3's minimum supported version (3.7)
warning: build failed, waiting for other jobs to finish... error: build failed error: cargo failed with code: 101
ERROR: Failed building wheel for tokenizers ERROR: Could not build wheels for tokenizers, which is required to install pyproject.toml-based projects
For python 3.7 and 3.8 i ran into another dependency issue for click:
Could not find a version that matches click==6.7,>=8.0
Hi,
I tried to test cde 2.1.2 lately on a Google Colab/Kaggle notebook without success. The installation of the library is troublesome and finish by the following error:
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. yellowbrick 1.4 requires scikit-learn>=1.0.0, but you have scikit-learn 0.22.1 which is incompatible. pip-tools 6.2.0 requires click>=7, but you have click 6.7 which is incompatible. imbalanced-learn 0.8.1 requires scikit-learn>=0.24, but you have scikit-learn 0.22.1 which is incompatible. google-colab 1.0.0 requires requests~=2.23.0, but you have requests 2.21.0 which is incompatible. en-core-web-sm 2.2.5 requires spacy>=2.2.2, but you have spacy 2.1.9 which is incompatible. datascience 0.10.6 requires folium==0.2.1, but you have folium 0.8.3 which is incompatible.
Python was 3.7.13 for Colab and 3.7.12 for Kaggle. I thought these versions were compatible with this version, is there something else that could cause a problem?
Thank you in advance!