2 layers for model loading

Dockerfile

@@ -1,6 +1,6 @@
 FROM python:3.11
 
-# Create and use a non-root user
+# Create and use a non-root user (optional)
 RUN useradd -m -u 1000 user
 USER user
 ENV PATH="/home/user/.local/bin:$PATH"
@@ -26,6 +26,9 @@ RUN mkdir -p $SENTENCE_TRANSFORMERS_HOME && \
 RUN python -c "from huggingface_hub import snapshot_download; \
     snapshot_download(repo_id='sentence-transformers/all-MiniLM-L6-v2', cache_dir='/app/model_cache')"
 
+# Ensure the model files are available at runtime (list out)
+RUN ls -l /app/model_cache && cat /app/model_cache/config.json
+
 # Ensure ownership and permissions remain intact
 RUN chown -R user:user /app/model_cache
 

app.py

@@ -60,10 +60,10 @@ os.environ["TOKENIZERS_PARALLELISM"] = "false"
 # os.environ["HF_HOME"] = huggingface_cache_dir  # Use this folder for HF cache
 # 2b) Setup Hugging Face Cloud project model cache
 hf_cache_dir = "/home/user/.cache/huggingface"
-os.environ["HF_HOME"] = hf_cache_dir
-os.environ["SENTENCE_TRANSFORMERS_HOME"] = hf_cache_dir
 # Model storage location
 model_cache_dir = "/app/model_cache"
+os.environ["HF_HOME"] = hf_cache_dir
+os.environ["SENTENCE_TRANSFORMERS_HOME"] = hf_cache_dir
 # 3. Download (or load from cache) the SentenceTransformer model 
 from huggingface_hub import snapshot_download
 print("⏳ Checking or downloading the all-MiniLM-L6-v2 model from huggingface_hub...")
@@ -74,22 +74,13 @@ if os.path.exists(model_cache_dir) and os.path.exists(os.path.join(model_cache_d
     model_loc = model_cache_dir
 # Else, try loading backup from snapshot_download
 else:
-    try:
-        model_loc = snapshot_download(
-            repo_id="sentence-transformers/all-MiniLM-L6-v2",
-            cache_dir=hf_cache_dir,
-            local_files_only=True  # ✅ Ensure it's loaded from cache
-        )
-        print(f"✅ Model loaded from local cache: {model_loc}")
-    except Exception as e:
-        print(f"❌ Error loading model from cache: {e}")
-        print("⚠️ Retrying with online download enabled...")
-        model_loc = snapshot_download(
-            repo_id="sentence-transformers/all-MiniLM-L6-v2",
-            cache_dir=hf_cache_dir,
-            local_files_only=False  # ✅ Fallback to online download
-        )
-        print(f"✅ Model directory after retry: {model_loc}")
+    print(f"❌ Model not found in {model_cache_dir}. This should not happen!")
+    print("⚠️ Retrying with snapshot_download...")
+    model_loc = snapshot_download(
+        repo_id="sentence-transformers/all-MiniLM-L6-v2",
+        cache_dir=hf_cache_dir,
+        local_files_only=True # Change to `False` for fallback to online download
+    )
 
 from sentence_transformers import SentenceTransformer
 print("📥 **Loading Embedding Model...**")