github/AMD-SHARK-Studio
1
1
Fork 0
mirror of https://github.com/nod-ai/AMD-SHARK-Studio.git synced 2026-04-03 03:00:17 -04:00
Code Issues Packages Projects Releases Wiki Activity

Make batch size configurable via pytest and fix sharktank generation. (#1227)

Browse Source 
* Fix sharktank generation and add batch_size pytest option for torch.

* Disable torch dynamo until py3.11 supported

* Compile torchmodel without dynamo if torch.compile fails

* Use release versions of TF/Keras for importer.

* Pin torchvision and remove debug prints.

* Remove duplicates from torch model list.

* Update generate_sharktank.py

* xfail a few models that fail sharktank generation/ numerics
This commit is contained in:
Ean Garvey
2023-03-28 14:33:39 -05:00
committed by GitHub
parent 0ef6a0e234
commit 96b1560da5
12 changed files with 244 additions and 129 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
.github/workflows/test-models.yml vendored
View File

@@ -120,7 +120,7 @@ jobs:
if: matrix.suite == 'cuda'
run: |
cd $GITHUB_WORKSPACE
PYTHON=python${{ matrix.python-version }} BENCHMARK=1 IMPORTER=1 ./setup_venv.sh
PYTHON=python${{ matrix.python-version }} ./setup_venv.sh
source shark.venv/bin/activate
pytest --forked --benchmark --ci --ci_sha=${SHORT_SHA} --update_tank --tank_url="gs://shark_tank/nightly/" -k cuda
gsutil cp ./bench_results.csv gs://shark-public/builder/bench_results/${DATE}/bench_results_cuda_${SHORT_SHA}.csv