rebase

Author: wwwjn

.github/workflows/integration_test_8gpu_flux.yaml

@@ -39,12 +39,12 @@ jobs:
         CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
         conda activate "${CONDA_ENV}"
 
+        pip config --user set global.progress_bar off
+
         # Log CUDA driver version for debugging.
         DRIVER_VERSION=$(nvidia-smi --query-gpu=driver_version --format=csv,noheader | head -n 1  || true)
         echo "CUDA driver version: ${DRIVER_VERSION}"
 
-        pip config --user set global.progress_bar off
-
         python -m pip install --force-reinstall --pre torch --index-url https://download.pytorch.org/whl/nightly/cu126
 
         mkdir artifacts-to-be-uploaded

torchtitan/experiments/flux/tests/integration_tests.py

@@ -86,8 +86,8 @@ def build_flux_test_list():
                     "--encoder.clip_encoder torchtitan/experiments/flux/tests/assets/clip-vit-large-patch14/",
                     "--encoder.t5_encoder torchtitan/experiments/flux/tests/assets/t5-v1_1-xxl/",
                     "--model.tokenizer_path tests/assets/tokenizer",
-                    "--parallelism.data_parallel_shard_degree=4",
-                    "--parallelism.data_parallel_replicate_degree=1",
+                    "--parallelism.data_parallel_shard_degree 4",
+                    "--parallelism.data_parallel_replicate_degree 1",
                 ]
             ],
             "FSDP",
@@ -102,14 +102,28 @@ def build_flux_test_list():
                     "--encoder.clip_encoder torchtitan/experiments/flux/tests/assets/clip-vit-large-patch14/",
                     "--encoder.t5_encoder torchtitan/experiments/flux/tests/assets/t5-v1_1-xxl/",
                     "--model.tokenizer_path tests/assets/tokenizer",
-                    "--parallelism.data_parallel_shard_degree=2",
-                    "--parallelism.data_parallel_replicate_degree=2",
+                    "--parallelism.data_parallel_shard_degree 2",
+                    "--parallelism.data_parallel_replicate_degree 2",
                 ]
             ],
             "HSDP",
             "hsdp",
             ngpu=4,
         ),
+        OverrideDefinitions(
+            [
+                [
+                    "--model.name flux",
+                    "--training.test_mode",
+                    "--encoder.clip_encoder torchtitan/experiments/flux/tests/assets/clip-vit-large-patch14/",
+                    "--encoder.t5_encoder torchtitan/experiments/flux/tests/assets/t5-v1_1-xxl/",
+                    "--model.tokenizer_path tests/assets/tokenizer",
+                    "--validation.enabled",
+                ]
+            ],
+            "Flux Validation Test",
+            "validation",
+        ),
     ]
     return integration_tests_flavors