cleanup

llama_stack/distribution/distribution.py

@@ -35,7 +35,6 @@ def builtin_automatically_routed_apis() -> List[AutoRoutedApiInfo]:
             routing_table_api=Api.memory_banks,
             router_api=Api.memory,
         ),
-        # TODO: re-enable once we have proper checking on builtin routing: Issue #297
         AutoRoutedApiInfo(
             routing_table_api=Api.datasets,
             router_api=Api.datasetio,

llama_stack/distribution/server/server.py

@@ -288,6 +288,7 @@ def main(
         apis_to_serve = set(impls.keys())
 
     for inf in builtin_automatically_routed_apis():
+        # if we do not serve the corresponding router API, we should not serve the routing table API
         if inf.router_api.value not in apis_to_serve:
             continue
         apis_to_serve.add(inf.routing_table_api.value)

tests/examples/local-run.yaml

@@ -13,18 +13,14 @@ apis:
 - safety
 providers:
   inference:
-  - provider_id: tgi0
+  - provider_id: meta-reference
-    provider_type: remote::tgi
+    provider_type: meta-reference
     config:
-      url: http://127.0.0.1:5009
+      model: Llama3.1-8B-Instruct
-  # - provider_id: meta-reference
+      quantization: null
-  #   provider_type: meta-reference
+      torch_seed: null
-  #   config:
+      max_seq_len: 4096
-  #     model: Llama3.1-8B-Instruct
+      max_batch_size: 1
-  #     quantization: null
-  #     torch_seed: null
-  #     max_seq_len: 4096
-  #     max_batch_size: 1
   safety:
   - provider_id: meta-reference
     provider_type: meta-reference