add commit_sha

Signed-off-by: Jared Wilber <jwilber@nvidia.com>

Author: jwilber

.github/workflows/convergence-tests.yml

@@ -59,7 +59,7 @@ jobs:
         run: |
           lep login -c "$LEP_LOGIN_CREDENTIALS" || true
           if [ -n "${{ inputs.config_override }}" ]; then
-            python ci/lepton/model_convergence/scripts/launch_job.py --config-name="${{ inputs.config_override }}.yaml" branch="${{ inputs.branch }}" commit_sha="${{ inputs.commit_sha }}"
+            python ci/lepton/model_convergence/scripts/launch_job.py --config-name="${{ inputs.config_override }}.yaml" branch="${{ inputs.branch }}" +commit_sha="${{ inputs.commit_sha }}"
           else
-            python ci/lepton/model_convergence/scripts/launch_job.py --config-name="recipes/${{ inputs.model_config }}" branch="${{ inputs.branch }}" commit_sha="${{ inputs.commit_sha }}"
+            python ci/lepton/model_convergence/scripts/launch_job.py --config-name="recipes/${{ inputs.model_config }}" branch="${{ inputs.branch }}" +commit_sha="${{ inputs.commit_sha }}"
           fi

ci/lepton/model_convergence/configs/recipes/amplify_accelerate_te_fp8.yaml

@@ -14,9 +14,12 @@ dashboard_info:
 # Base configuration
 recipe_subdir: amplify_accelerate_te_fp8
 branch: jwilber/add-wandb-recipe-amplify
+commit_sha: ""
 train_cmnd: train
 
 wandb_init_args:
+  project: "amplify_accelerate_te_fp8"
+  group: "recipes_model_convergence"
   mode: "online" # need online to collect logs; if offline, must manually sync after run
 
 stop_after_n_steps: 10
@@ -33,14 +36,20 @@ products:
 
 script: |
   git clone https://github.com/NVIDIA/bionemo-framework.git
-  if [ "${branch}" != "main" ]; then
-    cd bionemo-framework
+  cd bionemo-framework
+  if [ -n "${commit_sha}" ]; then
+    echo "Checking out commit: ${commit_sha}"
+    git checkout "${commit_sha}"
+  elif [ "${branch}" != "main" ]; then
+    echo "Checking out branch: ${branch}"
     git checkout "${branch}"
-    cd ..
   fi
+  cd ..
   cd bionemo-framework/recipes/${recipe_subdir}
   pip install -r requirements.txt
   torchrun ${train_cmnd}.py \
     --config-name ${config}.yaml \
-    stop_after_n_steps=${stop_after_n_steps} \
-    trainer.report_to=${trainer.report_to}
+    trainer.report_to=${trainer.report_to} \
+    wandb_init_args.mode=${wandb_init_args.mode} \
+    wandb_init_args.project=${wandb_init_args.project} \
+    +wandb_init_args.group=${wandb_init_args.group} \