Update app.py

app.py

@@ -458,6 +458,45 @@ def _install_supervisely_logger_shim():
     """))
     return str(root)
 
+# ---- NEW: robust fallback for cfg['_pymodule'] via sitecustomize -------------
+def _install_workspace_env_fallback(cwd_for_train: str, module_default: str = "rtdetrv2_pytorch.src"):
+    """
+    Creates a sitecustomize.py in the training cwd that monkeypatches
+    rtdetrv2_pytorch.src.core.workspace.create to gracefully handle missing
+    cfg['_pymodule'] by falling back to $RTDETR_PYMODULE or module_default.
+    """
+    sc_path = os.path.join(cwd_for_train, "sitecustomize.py")
+    code = textwrap.dedent(f"""
+        import os, importlib
+        try:
+            mod_path = os.environ.get("RTDETR_PYMODULE", "{module_default}")
+            ws_mod = importlib.import_module("rtdetrv2_pytorch.src.core.workspace")
+            _orig_create = ws_mod.create
+            def _safe_create(name, cfg, *args, **kwargs):
+                pm = None
+                try:
+                    pm = cfg.get("_pymodule", None)
+                except Exception:
+                    pm = None
+                if not pm:
+                    pm = os.environ.get("RTDETR_PYMODULE", "{module_default}")
+                try:
+                    importlib.import_module(pm)
+                except Exception:
+                    pass
+                try:
+                    cfg["_pymodule"] = pm
+                except Exception:
+                    pass
+                return _orig_create(name, cfg, *args, **kwargs)
+            ws_mod.create = _safe_create
+        except Exception:
+            pass
+    """)
+    with open(sc_path, "w", encoding="utf-8") as f:
+        f.write(code)
+    return sc_path
+
 def _ensure_checkpoint(model_key: str, out_dir: str) -> str | None:
     url = CKPT_URLS.get(model_key)
     if not url:
@@ -571,6 +610,9 @@ def patch_base_config(base_cfg_path, merged_dir, class_count, run_name,
 
     # Disable SyncBN for single GPU/CPU runs
     cfg["sync_bn"] = False
+    # Guardrails for single-process runs
+    cfg.setdefault("device", "")
+    cfg["find_unused_parameters"] = False
 
     # Remove COCO dataset include so it can't override our dataset paths later
     if "__include__" in cfg and isinstance(cfg["__include__"], list):
@@ -825,6 +867,10 @@ def training_handler(dataset_path, model_key, run_name, epochs, batch, imgsz, lr
     q = Queue()
     def run_train():
         try:
+            # Ensure our fallback hook is available in the train process (CWD on sys.path)
+            train_cwd = os.path.dirname(train_script)
+            _install_workspace_env_fallback(train_cwd)
+
             env = os.environ.copy()
             # Make sure repo code can be imported
             env["PYTHONPATH"] = os.pathsep.join(filter(None, [
@@ -837,7 +883,7 @@ def training_handler(dataset_path, model_key, run_name, epochs, batch, imgsz, lr
             # Provide a secondary hint for some config loaders
             env.setdefault("RTDETR_PYMODULE", "rtdetrv2_pytorch.src")
 
-            proc = subprocess.Popen(cmd, cwd=os.path.dirname(train_script),
+            proc = subprocess.Popen(cmd, cwd=train_cwd,
                                     stdout=subprocess.PIPE, stderr=subprocess.STDOUT,
                                     bufsize=1, text=True, env=env)
             for line in proc.stdout: