Use cpu wheel for torchvision (#155)

Making it compatible with the installed torch
version. Addresses build failures of the kind:

RuntimeError: operator torchvision::nms does not exist

Also, remove unnecessary s3torchconnector version
constraint.

Co-authored-by: Vishal Gowda <[email protected]>

Author: vishalg0wda

.github/workflows/python-checks.yml

@@ -48,7 +48,7 @@ jobs:
         run: |
           python -m pip install --upgrade pip
           # Manually install CPU-only version of torch so we're not carrying around giant GPU drivers/kernels
-          python -m pip install torch --extra-index-url https://download.pytorch.org/whl/cpu
+          python -m pip install torch torchvision --extra-index-url https://download.pytorch.org/whl/cpu
           python -m pip install -e "s3torchconnectorclient[test]"
           python -m pip install -e "s3torchconnector[test]"
           python -m pip install -e "s3torchbenchmarking[test]"

s3torchbenchmarking/pyproject.toml

@@ -10,7 +10,7 @@ requires-python = ">=3.8,<3.12"
 readme = "README.md"
 dependencies = [
     "torch >= 2.0.1",
-    "s3torchconnector>=1.1.1",
+    "s3torchconnector",
     "hydra-core",
     "torchdata>=0.6.1",
     "torchvision",