Python InterOp - Tuples in API calls - Pytorch example

[diro5t]

Hi,

first of all: I am a huge fan of having a lispy language hosted on Python with access to all libraries - FANTASTIC!!!

I have a little issue: I don't know how to express an int tuple as an argument in a Python API call

Please help ...

EXAMPLE:

(import torch)
;; torch.rand expects int tuple - works
(def t4 (torch/rand 2 2))
(println t4)
,, torch.reshape expects tensor plus int tuple - crashes
(def t5 (torch/reshape t4 2 2))
(println t5)

[chrisrink10]

Hi @diro5t it's great to hear you like Basilisp. In the future for questions, can you use the discussions section rather than filing an issue?

As to your request, Basilisp has built in data readers for Python primitive types:

#py (2 2)

I am not familiar with the particulars of Pytorch, so I do not know why certain calls are failing. I'm guessing this has to do with how the Python API is defined (e.g. supporting variadic arguments versus expecting a tuple argument directly).

[diro5t]

I am so sorry. I didn't see the section in the docs. Thanks so much! Again, this project is amazing and deserves a much broader public footprint. A homoiconic language with macros and full access to Python libraries - this is so helpful for people doing DSL, data science and AI. There is even a Jupyter Kernel. Keep up the good work!