Make torch 2.14 the unit-test default and constrain it for the tensorrt example images (#2309)

### What does this PR do?

Type of change: Bug fix (CI) + test coverage

**Fixes `onnx (torch_onnx)` and `onnx (diffusers)`**, which have failed
on every branch since
`torch 2.14.0` was published to PyPI today (2026-09-02 13:42 UTC), and
**adds torch 2.14 to the unit
test matrix as the new default** so the next torch release is caught
there rather than in an example job.

### Root cause

Every test in those two jobs failed with:

```
RuntimeError: CUDNN_BACKEND_TENSOR_DESCRIPTOR cudnnFinalize failed
  ptrDesc->finalize() cudnn_status: CUDNN_STATUS_SUBLIBRARY_LOADING_FAILED
```

`nvcr.io/nvidia/tensorrt:26.05-py3` ships cuDNN **9.22** and has no
preinstalled torch, so pip
resolved the newest one — and torch 2.14 pins
`nvidia-cudnn-cu13==9.24.0.43`. Loading 9.24
sublibraries against the image's 9.22 `libcudnn.so.9` is exactly what
that status reports.

| | last good run (08:55) | first failing run (13:34) |
|---|---|---|
| `torch` | 2.13.0 | **2.14.0** |
| `nvidia-cudnn-cu13` | 9.20.0.48 | **9.24.0.43** |
| image cuDNN | 9.22.0.52 | 9.22.0.52 |

### Why only these two jobs

- The **nemo** and **pytorch** images have a preinstalled torch that
already satisfies `torch>=2.8`,
so pip never resolves a new one — confirmed from the megatron job log,
where torch does not appear
  in `Successfully installed`.
- **`tensorrt:26.05-py3` has no preinstalled torch**, so pip takes the
newest from PyPI.
- **`onnx (torch_trt)`** shares that image but passes throughout,
because `torch-tensorrt<2.13`
  already holds torch below 2.14.

### The changes

1. **Constrain torch only where the incompatibility is.**
`PIP_CONSTRAINT=torch<2.14` in the example
runner, applied when the job's image is a `tensorrt` one. It also covers
the
`examples/*/requirements.txt` loop in the same shell, which matters
because `nemo_automodel`
pulls torch in too. Not pinned in `pyproject.toml`: torch 2.14 is fine
anywhere its own bundled
cuDNN is the one loaded, so that would constrain users to work around
one pinned image.
2. **Test torch 2.14.** `torch_214` added to `TORCH_VERSIONS`
(`torchvision~=0.29.0`) and promoted to
the unit-test default across the supported Python versions, with 2.13
demoted to the back-compat
row. `release.yml`'s basic unit test moves to the same default (it was
still on 2.12).
Nothing exercised 2.14 before — which is why a torch release reached us
through an example
   job instead of a unit test.

### Testing

- `actionlint` and YAML/TOML parse clean; pre-commit clean.
- Verified by this PR's own jobs: `onnx (torch_onnx)` and `onnx
(diffusers)` reproduce the failure on
`main` right now, and the new `unit-3.12(torch_214, tf_latest)` job is
the first run of ModelOpt
  against torch 2.14.

### Before your PR is "*Ready for review*"

- Is this change backward compatible?: ✅ — CI-only; no source or package
metadata change
- If you copied code from any other sources or added a new PIP
dependency, did you follow guidance in `CONTRIBUTING.md`: N/A — no new
dependency
- Did you write any new necessary tests?: ✅ — torch 2.14 added to the
unit test matrix
- Did you update
[Changelog](https://github.com/NVIDIA/Model-Optimizer/blob/main/CHANGELOG.rst)?:
N/A — internal CI, not user-facing
- Did you get Claude approval on this PR?: ❌ — not yet requested

---------

Signed-off-by: Keval Morabia <28916987+kevalmorabia97@users.noreply.github.com>

.github/workflows/_example_tests_runner.yml

@@ -62,6 +62,13 @@ jobs:
           # Uninstall conflicting system-wide installed modelopt in nemo containers
           pip uninstall -y nvidia-modelopt || true
 
+          # nvcr.io/nvidia/tensorrt:26.05-py3 ships cuDNN 9.22 with no preinstalled torch, and
+          # torch 2.14 pins cuDNN 9.24: mixing them fails with CUDNN_SUBLIBRARY_LOADING_FAILED.
+          if [[ "${{ inputs.docker_image }}" == *"/tensorrt:"* ]]; then
+            echo "torch<2.14" > /tmp/pip-constraints.txt
+            export PIP_CONSTRAINT=/tmp/pip-constraints.txt
+          fi
+
           # Use `python -m pip` instead of `pip` to avoid conflicts with system pip for nemo containers
           # Editable install so example scripts launched as subprocesses resolve modelopt to the same source path as the test process
           python -m pip install -e ".${{ inputs.pip_install_extras }}"

.github/workflows/code_quality.yml

@@ -29,4 +29,10 @@ jobs:
           # verified. Only nightly sees them, since it scans all history while PRs scan their diff.
           extra_args: --results=verified,unknown --exclude-detectors=lob
       - name: Run code quality checks
-        run: pip install nox uv && nox -s pre_commit_all
+        run: |
+          # torch 2.14 breaks this env: torchvision::nms fails to register, which disables
+          # modelopt's transformers plugin and takes the ARGUMENTS.md hook down with it.
+          echo "torch<2.14" > /tmp/pip-constraints.txt
+          # Both: nox builds the session venv with uv, which reads UV_CONSTRAINT, not PIP_CONSTRAINT.
+          export PIP_CONSTRAINT=/tmp/pip-constraints.txt UV_CONSTRAINT=/tmp/pip-constraints.txt
+          pip install nox uv && nox -s pre_commit_all

.github/workflows/release.yml

@@ -14,7 +14,7 @@ jobs:
       - uses: actions/checkout@v6
       - uses: ./.github/actions/ubuntu-setup
       - name: Run basic unit tests
-        run: pip install nox uv && nox -s "unit-3.12(torch_212, tf_latest)"
+        run: pip install nox uv && nox -s "unit-3.12(torch_214, tf_latest)"
       - name: Build Wheel
         run: |
           nox -s build_wheel

.github/workflows/unit_tests.yml

@@ -67,7 +67,7 @@ jobs:
         env:
           COVERAGE_PROCESS_START: ${{ github.workspace }}/pyproject.toml
           COVERAGE_FILE: ${{ github.workspace }}/.coverage
-        run: pip install nox uv && nox -s "unit-3.12(torch_213, tf_latest)"
+        run: pip install nox uv && nox -s "unit-3.12(torch_214, tf_latest)"
       - name: Upload coverage reports to Codecov
         uses: codecov/codecov-action@v7
         with:
@@ -92,7 +92,7 @@ jobs:
         with:
           python-version: "3.12"
       - name: Run unit tests (without coverage)
-        run: pip install nox uv && nox -s "unit-3.12(torch_213, tf_latest)"
+        run: pip install nox uv && nox -s "unit-3.12(torch_214, tf_latest)"
   multi-version:
     if: needs.check-file-changes.outputs.any_changed == 'true'
     needs: [linux, check-file-changes]
@@ -102,19 +102,20 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          # Default torch (2.13) across the other supported Python versions
-          - {nox_session: "unit-3.10(torch_213, tf_latest)", python_version: "3.10"}
-          - {nox_session: "unit-3.11(torch_213, tf_latest)", python_version: "3.11"}
-          - {nox_session: "unit-3.13(torch_213, tf_latest)", python_version: "3.13"}
-          - {nox_session: "unit-3.14(torch_213, tf_latest)", python_version: "3.14"}
+          # Default torch (2.14) across the other supported Python versions
+          - {nox_session: "unit-3.10(torch_214, tf_latest)", python_version: "3.10"}
+          - {nox_session: "unit-3.11(torch_214, tf_latest)", python_version: "3.11"}
+          - {nox_session: "unit-3.13(torch_214, tf_latest)", python_version: "3.13"}
+          - {nox_session: "unit-3.14(torch_214, tf_latest)", python_version: "3.14"}
           # Older torch versions on the default Python (3.12) for back-compat.
           - {nox_session: "unit-3.12(torch_28, tf_latest)", python_version: "3.12"}
           - {nox_session: "unit-3.12(torch_29, tf_latest)", python_version: "3.12"}
           - {nox_session: "unit-3.12(torch_210, tf_latest)", python_version: "3.12"}
           - {nox_session: "unit-3.12(torch_211, tf_latest)", python_version: "3.12"}
           - {nox_session: "unit-3.12(torch_212, tf_latest)", python_version: "3.12"}
+          - {nox_session: "unit-3.12(torch_213, tf_latest)", python_version: "3.12"}
           # Minimum supported transformers on the default torch.
-          - {nox_session: "unit-3.12(torch_213, tf_min)", python_version: "3.12"}
+          - {nox_session: "unit-3.12(torch_214, tf_min)", python_version: "3.12"}
     steps:
       - uses: actions/checkout@v6
       - uses: ./.github/actions/ubuntu-setup

noxfile.py

@@ -40,6 +40,7 @@ TORCH_VERSIONS = {
     "torch_211": "torchvision~=0.26.0",
     "torch_212": "torchvision~=0.27.0",
     "torch_213": "torchvision~=0.28.0",
+    "torch_214": "torchvision~=0.29.0",
 }
 
 # Extra install pins applied per transformers matrix entry (installed after the base