NeuromatchAcademy / course-content

NMA Computational Neuroscience course

Home Page:https://compneuro.neuromatch.io

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

requirements.txt

arash-ashra opened this issue · comments

I was wondering if the requirements.txt tested on all the notebooks? it looks pretty short and versions are not included

I am planning on creating a virtual environment for offline access for all the students using this.

It is what's used for the automated CI, so yes.

It was a conscious decision to keep the set of required libraries small and limited to core scientific Python libraries.

This is actually a subset of the libraries used across the tutorials; there are a few one-off days that use more domain-specific libraries (i.e. the HMM library used for stochastic processes), but the rule is that tutorials using libraries not defined in this requirements set have to pip install them within the notebook itself, and using more esoteric libraries was generally discouraged during content creation.

There are no versions because there is not, AFAIK, a public record of the versions installed on Colab, so it seemed better to test against "latest versions" rather than pin on versions that will fall out of sync with Colab.

Thanks Michael for the clarification. I'll find a workaround.

May be useful:

[mlw-mbpro tutorials]{master $}$ git grep '!pip'
Bonus_UnsupervisedLearning/Tutorial2.ipynb:    "!pip install plotly --quiet\n",
W0D4_Calculus/W0D4_Tutorial1.ipynb:    "!pip install sympy --quiet\n",
W3D2_HiddenDynamics/W3D2_Tutorial2.ipynb:    "!pip install hmmlearn --quiet\n",
W3D2_HiddenDynamics/W3D2_Tutorial4.ipynb:    "!pip install pykalman --quiet\n",
W3D5_NetworkCausality/W3D5_Tutorial4.ipynb:    "!pip install statsmodels\n",