Benchmarks: enable W&B reporting on officeqa, swe-atlas-qna, and tau3

varunursekar · varunursekar · commit c89ef8429aa4 · 2026-07-23T23:16:55.000-07:00
diff --git a/harness-engineering-bench/officeqa/baseline/build.yaml b/harness-engineering-bench/officeqa/baseline/build.yaml
@@ -73,6 +73,14 @@ verifier_timeout_seconds: 3600
 secrets:
   - MODAL_TOKEN_ID
   - MODAL_TOKEN_SECRET
+  - WANDB_API_KEY
+  - WANDB_BASE_URL  # self-hosted W&B (scaleai.wandb.io)
+
+wandb:
+  project: vero-officeqa
+  group: officeqa
+  log_traces: true
+
 inference_gateway:
   upstream_api_key_env: OPENAI_API_KEY
   upstream_base_url_env: OPENAI_BASE_URL
diff --git a/harness-engineering-bench/swe-atlas-qna/baseline/build.yaml b/harness-engineering-bench/swe-atlas-qna/baseline/build.yaml
@@ -65,6 +65,14 @@ verifier_timeout_seconds: 14400
 secrets:
   - MODAL_TOKEN_ID
   - MODAL_TOKEN_SECRET
+  - WANDB_API_KEY
+  - WANDB_BASE_URL  # self-hosted W&B (scaleai.wandb.io)
+
+wandb:
+  project: vero-swe-atlas-qna
+  group: swe-atlas-qna
+  log_traces: true
+
 inference_gateway:
   upstream_api_key_env: OPENAI_API_KEY
   upstream_base_url_env: OPENAI_BASE_URL
diff --git a/harness-engineering-bench/tau3/baseline/build.yaml b/harness-engineering-bench/tau3/baseline/build.yaml
@@ -64,6 +64,14 @@ verifier_timeout_seconds: 14400
 secrets:
   - MODAL_TOKEN_ID
   - MODAL_TOKEN_SECRET
+  - WANDB_API_KEY
+  - WANDB_BASE_URL  # self-hosted W&B (scaleai.wandb.io)
+
+wandb:
+  project: vero-tau3
+  group: tau3
+  log_traces: true
+
 inference_gateway:
   upstream_api_key_env: OPENAI_API_KEY
   upstream_base_url_env: OPENAI_BASE_URL
