fix: pre-download SAP model weights and unify HF login

Dockerfile

@@ -37,6 +37,10 @@ RUN pip install --no-cache-dir -r webapp/requirements.txt
 # Install SAP-RPT-1 OSS directly from GitHub (needed for the real model)
 RUN pip install --no-cache-dir git+https://github.com/SAP-samples/sap-rpt-1-oss.git
 
+# Pre-download Sentence Transformers weights to avoid runtime hangs
+# This specific model is used by the SAP RPT-1 OSS model.
+RUN python -c "from sentence_transformers import SentenceTransformer; SentenceTransformer('all-MiniLM-L6-v2')"
+
 # Expose port 7860 (Hugging Face Spaces default port)
 EXPOSE 7860
 

webapp/benchmark.py

@@ -88,8 +88,6 @@ class _SAPModel:
                     self._model = _SAPModel._shared_model._model
                     self._real  = True
                 else:
-                    from huggingface_hub import login
-                    login(token=HF_TOKEN, add_to_git_credential=False)
                     from sap_rpt_oss import SAP_RPT_OSS_Classifier, SAP_RPT_OSS_Regressor
                     if task == "classification":
                         self._model = SAP_RPT_OSS_Classifier(max_context_size=2048, bagging=1)

webapp/main.py

@@ -21,8 +21,17 @@ BASE_DIR = Path(__file__).resolve().parent.parent
 load_dotenv(BASE_DIR / "webapp" / ".env")
 
 # Verify Secrets on startup
+hf_token = os.environ.get('HUGGING_FACE_HUB_TOKEN')
 logger.info(f"TABPFN_TOKEN status: {'SET' if os.environ.get('TABPFN_TOKEN') else 'MISSING'}")
-logger.info(f"HF_TOKEN status:     {'SET' if os.environ.get('HUGGING_FACE_HUB_TOKEN') else 'MISSING'}")
+logger.info(f"HF_TOKEN status:     {'SET' if hf_token else 'MISSING'}")
+
+if hf_token:
+    try:
+        from huggingface_hub import login
+        login(token=hf_token, add_to_git_credential=False)
+        logger.info("Successfully logged into Hugging Face Hub.")
+    except Exception as e:
+        logger.warning(f"Hugging Face login failed: {e}")
 
 # ── Config ─────────────────────────────────────────────────────────────────────
 MAX_FILE_BYTES = int(os.getenv("MAX_FILE_SIZE_MB", "5")) * 1024 * 1024