Upload folder using huggingface_hub

.env.example

@@ -9,4 +9,5 @@ VESPA_CLOUD_MTLS_KEY="-----BEGIN PRIVATE KEY-----
 -----END PRIVATE KEY-----"
 VESPA_CLOUD_MTLS_CERT="-----BEGIN CERTIFICATE-----
 ...
------END CERTIFICATE-----"
+-----END CERTIFICATE-----"
+HOT_RELOAD=true

backend/colpali.py

@@ -309,6 +309,8 @@ def add_sim_maps_to_result(
         vespa_sim_map = single_result["fields"].get("summaryfeatures", None)
         if vespa_sim_map:
             vespa_sim_maps.append(vespa_sim_map)
+    if not imgs:
+        return result
     sim_map_imgs_generator = gen_similarity_maps(
         model=model,
         processor=processor,
@@ -322,7 +324,14 @@ def add_sim_maps_to_result(
     )
     for img_idx, token, sim_mapb64 in sim_map_imgs_generator:
         print(f"Created sim map for image {img_idx} and token {token}")
-        result["root"]["children"][img_idx]["fields"][f"sim_map_{token}"] = sim_mapb64
+        if (
+            len(result["root"]["children"]) > img_idx
+            and "fields" in result["root"]["children"][img_idx]
+            and "sim_map" in result["root"]["children"][img_idx]["fields"]
+        ):
+            result["root"]["children"][img_idx]["fields"][f"sim_map_{token}"] = (
+                sim_mapb64
+            )
         # Update result_cache with the new sim_map
         result_cache.set(query_id, result)
     # for single_result, sim_map_dict in zip(result["root"]["children"], sim_map_imgs):

backend/vespa_app.py

@@ -279,6 +279,9 @@ class VespaQueryClient:
             raise ValueError(f"Unsupported ranking: {ranking}")
 
         # Print score, title id, and text of the results
+        if "root" not in result or "children" not in result["root"]:
+            result["root"] = {"children": []}
+            return result
         for idx, child in enumerate(result["root"]["children"]):
             print(
                 f"Result {idx+1}: {child['relevance']}, {child['fields']['title']}, {child['fields']['id']}"

main.py

@@ -1,12 +1,12 @@
 import asyncio
 import base64
-import hashlib
 import io
 import os
 import time
 from concurrent.futures import ThreadPoolExecutor
 from functools import partial
 from pathlib import Path
+import uuid
 
 import google.generativeai as genai
 from fasthtml.common import *
@@ -112,7 +112,7 @@ async def keepalive():
 
 
 def generate_query_id(query):
-    return hashlib.md5(query.encode("utf-8")).hexdigest()
+    return uuid.uuid4().hex
 
 
 @rt("/static/{filepath:path}")
@@ -394,4 +394,6 @@ def get():
 
 if __name__ == "__main__":
     # ModelManager.get_instance()  # Initialize once at startup
-    serve(port=7860, reload=False)
+    HOT_RELOAD = os.getenv("HOT_RELOAD", "False").lower() == "true"
+    print(f"Starting app with hot reload: {HOT_RELOAD}")
+    serve(port=7860, reload=HOT_RELOAD)