Ship only the TensorRT delegate in the ExecuTorch runtime wheel - #4567

Open

shoumikhin wants to merge 17 commits into

pytorch:mainfrom

shoumikhin:executorch-slim-runtime-wheel

Commits on Aug 25, 2026

ci(executorch): run a coalesced TensorRT + CUDA program in the reference runner gate
shoumikhin
committed
10c8110
Track a recent ExecuTorch pin instead of 1.4.1
shoumikhin
committed
0c74fc9
Install the ExecuTorch the pin names, from the channel that has it
shoumikhin
committed
ba00b25
Follow the row's CUDA version, and reach the channel from every example
shoumikhin
committed
fccbf67
Keep the extra resolvable on Windows, and pin the channel docgen installs from
shoumikhin
committed
c4286aa
Put the index on the two printed installs, and fail the suite when setup fails
shoumikhin
committed
8c3bb6e
Assert the pin sites CI actually uses, and move the checks off the GPU lane
shoumikhin
committed
6f78926
Make the pin checks run in CI, and test the properties they assert
shoumikhin
committed
894d647
Let the pairing check run on the one lane that installs ExecuTorch
shoumikhin
committed
76c3459
Make the pin guards read values instead of nearby text
shoumikhin
committed
9e0f4b9
Stop the lockfile check from going red repo-wide on the next refresh
shoumikhin
committed
39c2552
Stop a test executing Python read out of a workflow file
shoumikhin
committed
b4d9945
Guard the nightly channel at the local-path install sites, and check it resolves
shoumikhin
committed
97b636b
Harden the ExecuTorch pin guards and the printed install commands
shoumikhin
committed
bba6c9b

Commits on Aug 26, 2026