training + orchestration console logs (only copy — no hub duplicate existed)

logs/attach_dual.log

@@ -0,0 +1,14 @@
+=== [0] gate ===
+0, NVIDIA A40, 46068 MiB
+1, NVIDIA A40, 46068 MiB
+both GPUs compute OK
+RAM cgroup: ?
+mfs#ca-mtl-1.runpod.net:9421  965T  725T  241T  76% /workspace
+=== [1] deps + tools ===
+[notice] To update, run: python -m pip install --upgrade pip
+  fetched captionbert_8192_v2.py
+  fetched ar_differentiation_bed.py
+  fetched geolip_vitals.py
+  fetched alephlm_0.py
+=== [2] parallel lanes ===
+lane0 pid 1883 (a3-s1, +75min stagger) | lane1 pid 1884 (a1-s1 resume -> a2-s1)