diff --git a/.github/workflows/semantic-pr.yml b/.github/workflows/semantic-pr.yml
index 4df7324c4..57a4df646 100644
--- a/.github/workflows/semantic-pr.yml
+++ b/.github/workflows/semantic-pr.yml
@@ -11,7 +11,7 @@ on:
       - synchronize
 
 concurrency:
-  group: ${{ github.workflow }}-${{ github.ref }}
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number }}
   cancel-in-progress: true
 
 permissions:
diff --git a/llama_stack/distributions/dell/dell.py b/llama_stack/distributions/dell/dell.py
index b561ea00e..e3bf0ee03 100644
--- a/llama_stack/distributions/dell/dell.py
+++ b/llama_stack/distributions/dell/dell.py
@@ -16,6 +16,7 @@ from llama_stack.distributions.template import DistributionTemplate, RunConfigSe
 from llama_stack.providers.inline.inference.sentence_transformers import (
     SentenceTransformersInferenceConfig,
 )
+from llama_stack.providers.remote.vector_io.chroma import ChromaVectorIOConfig
 
 
 def get_distribution_template() -> DistributionTemplate:
@@ -71,9 +72,10 @@ def get_distribution_template() -> DistributionTemplate:
     chromadb_provider = Provider(
         provider_id="chromadb",
         provider_type="remote::chromadb",
-        config={
-            "url": "${env.CHROMA_URL}",
-        },
+        config=ChromaVectorIOConfig.sample_run_config(
+            f"~/.llama/distributions/{name}/",
+            url="${env.CHROMADB_URL:=}",
+        ),
     )
 
     inference_model = ModelInput(
diff --git a/llama_stack/distributions/dell/run-with-safety.yaml b/llama_stack/distributions/dell/run-with-safety.yaml
index ecc6729eb..d89c92aa1 100644
--- a/llama_stack/distributions/dell/run-with-safety.yaml
+++ b/llama_stack/distributions/dell/run-with-safety.yaml
@@ -26,7 +26,10 @@ providers:
   - provider_id: chromadb
     provider_type: remote::chromadb
     config:
-      url: ${env.CHROMA_URL}
+      url: ${env.CHROMADB_URL:=}
+      kvstore:
+        type: sqlite
+        db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/dell/}/chroma_remote_registry.db
   safety:
   - provider_id: llama-guard
     provider_type: inline::llama-guard
diff --git a/llama_stack/distributions/dell/run.yaml b/llama_stack/distributions/dell/run.yaml
index fc2553526..7397410ba 100644
--- a/llama_stack/distributions/dell/run.yaml
+++ b/llama_stack/distributions/dell/run.yaml
@@ -22,7 +22,10 @@ providers:
   - provider_id: chromadb
     provider_type: remote::chromadb
     config:
-      url: ${env.CHROMA_URL}
+      url: ${env.CHROMADB_URL:=}
+      kvstore:
+        type: sqlite
+        db_path: ${env.SQLITE_STORE_DIR:=~/.llama/distributions/dell/}/chroma_remote_registry.db
   safety:
   - provider_id: llama-guard
     provider_type: inline::llama-guard
diff --git a/llama_stack/providers/remote/inference/tgi/tgi.py b/llama_stack/providers/remote/inference/tgi/tgi.py
index a5bb079ef..323831845 100644
--- a/llama_stack/providers/remote/inference/tgi/tgi.py
+++ b/llama_stack/providers/remote/inference/tgi/tgi.py
@@ -308,9 +308,7 @@ class TGIAdapter(_HfAdapter):
         if not config.url:
             raise ValueError("You must provide a URL in run.yaml (or via the TGI_URL environment variable) to use TGI.")
         log.info(f"Initializing TGI client with url={config.url}")
-        self.client = AsyncInferenceClient(
-            model=config.url,
-        )
+        self.client = AsyncInferenceClient(model=config.url, provider="hf-inference")
         endpoint_info = await self.client.get_endpoint_info()
         self.max_tokens = endpoint_info["max_total_tokens"]
         self.model_id = endpoint_info["model_id"]