Migrate from LLama Stack to OGX [main]

.tekton/integration-tests/pipeline/rag-content-0-8-integration-test.yaml

@@ -466,8 +466,6 @@ spec:
                 echo "CDI devices:"
                 nvidia-ctk cdi list || true
               fi
-              # Install required test deps
-              sudo dnf -y install python3 jq diffutils
               cd $HOME/rag-content
               SCRIPTEOF
               # Inject env vars

src/lightspeed_rag_content/config_templates.py

@@ -21,10 +21,9 @@
 image_name: starter
 
 apis:
-- agents
 - files
 - inference
-- safety
+- responses
 - tool_runtime
 - vector_io
 
@@ -44,21 +43,10 @@
       storage_dir: /tmp/files
     provider_id: meta-reference-files
     provider_type: inline::localfs
-  agents:
-  - config:
-      persistence:
-        agent_state:
-          namespace: agents_state
-          backend: kv_default
-        responses:
-          table_name: agents_responses
-          backend: sql_default
-    provider_id: meta-reference
-    provider_type: inline::meta-reference
   tool_runtime:
   - config: {{}}
-    provider_id: rag-runtime
-    provider_type: inline::rag-runtime
+    provider_id: file-search
+    provider_type: inline::file-search
   vector_io:
   - config:
       {vector_io_cfg}
@@ -99,13 +87,13 @@
   scoring_fns: []
   benchmarks: []
   tool_groups:
-  - toolgroup_id: builtin::rag
-    provider_id: rag-runtime
+  - toolgroup_id: builtin::file-search
+    provider_id: file-search
 """
 
 OGX_VECTOR_STORES_TEMPLATE = """vector_stores:
   - embedding_dimension: {dimension}
-    embedding_model: sentence-transformers/{model_name_or_dir}
+    embedding_model: {model_name}
     provider_id: {vector_io_provider_id}
     vector_store_id: {vector_store_id}"""
 
@@ -169,8 +157,8 @@
     db_path: ${{env.RAG_DB_PATH:={db_path}}}
 
 rag:
-  # inline:
-  #   - {index_id}
+  inline:
+    - {index_id}
   tool:
     - {index_id}
 """
@@ -189,8 +177,8 @@
     password: ${{env.POSTGRES_PASSWORD}}
 
 rag:
-  # inline:
-  #   - {index_id}
+  inline:
+    - {index_id}
   tool:
     - {index_id}
 """

src/lightspeed_rag_content/document_processor.py

@@ -489,7 +489,7 @@ async def _insert_prechunked_documents(  # pylint: disable=R0914
             name=index,
             extra_body={
                 "provider_id": index,
-                "embedding_model": f"sentence-transformers/{self.model_name_or_dir}",
+                "embedding_model": self.config.model_name,
                 "embedding_dimension": self.config.embedding_dimension,
             },
         )
@@ -543,7 +543,7 @@ async def upload_file(chunk_indices: list[int]) -> str:
         # Compute embeddings
         LOG.info("Computing embeddings for %d chunks...", len(self.documents))
         chunks_to_insert = []
-        embedding_model = f"sentence-transformers/{self.model_name_or_dir}"
+        embedding_model = self.config.model_name
 
         for idx, doc in enumerate(self.documents):
             if isinstance(doc, dict):
@@ -587,7 +587,7 @@ async def _upload_and_process_files(  # noqa: C901  # pylint: disable=R0912,R091
             name=index,
             extra_body={
                 "provider_id": index,
-                "embedding_model": f"sentence-transformers/{self.model_name_or_dir}",
+                "embedding_model": self.config.model_name,
                 "embedding_dimension": self.config.embedding_dimension,
             },
         )
@@ -726,7 +726,7 @@ def _update_yaml_config(self, cfg_file: str, index: str, vector_store_id: str) -
         """Update the config file with the created vector_store_id."""
         vector_stores_section = self.VECTOR_STORES_TEMPLATE.format(
             dimension=self.config.embedding_dimension,
-            model_name_or_dir=self.model_name_or_dir,
+            model_name=self.config.model_name,
             vector_io_provider_id=index,
             vector_store_id=vector_store_id,
         )

tests/integration-konflux/config/run.yaml

@@ -2,10 +2,9 @@ version: 2
 image_name: starter
 
 apis:
-- agents
 - files
 - inference
-- safety
+- responses
 - tool_runtime
 - vector_io
 
@@ -27,26 +26,18 @@ providers:
       storage_dir: /tmp/files
     provider_id: meta-reference-files
     provider_type: inline::localfs
-  safety:
-  - config:
-      excluded_categories: []
-    provider_id: llama-guard
-    provider_type: inline::llama-guard
-  agents:
+  responses:
   - config:
       persistence:
-        agent_state:
-          namespace: agents_state
-          backend: kv_default
         responses:
-          table_name: agents_responses
           backend: sql_default
-    provider_id: meta-reference
-    provider_type: inline::meta-reference
+          table_name: responses
+    provider_id: builtin-responses
+    provider_type: inline::builtin
   tool_runtime:
   - config: {}
-    provider_id: rag-runtime
-    provider_type: inline::rag-runtime
+    provider_id: file-search
+    provider_type: inline::file-search
   vector_io: []
 
 server:
@@ -70,6 +61,9 @@ storage:
     conversations:
       table_name: openai_conversations
       backend: sql_default
+    responses:
+      table_name: responses
+      backend: sql_default
 
 registered_resources:
   models:
@@ -79,17 +73,12 @@ registered_resources:
     provider_model_id: /embeddings
     metadata:
       embedding_dimension: 768
-  shields:
-  - shield_id: llama-guard
-    provider_id: llama-guard
-    provider_shield_id: openai/gpt-4o-mini
+  shields: []
   vector_stores: []
   tool_groups:
-  - toolgroup_id: builtin::rag
-    provider_id: rag-runtime
+  - toolgroup_id: builtin::file-search
+    provider_id: file-search
 
-safety:
-  default_shield_id: llama-guard
 vector_stores:
   default_provider_id: faiss
   default_embedding_model:

tests/integration-konflux/pipeline-konflux.sh

@@ -133,6 +133,7 @@ podman run -d --name lightspeed-stack-e2e \
   -e OPENAI_API_KEY="$OPENAI_API_KEY" \
   -e RAG_DB_PATH=/opt/app-root/src/.llama/storage/rag/faiss_store.db \
   -e HF_HOME=/embeddings \
+  -e OTEL_SDK_DISABLED=true \
   ${HF_TOKEN:+-e HF_TOKEN="$HF_TOKEN"} \
   "$LIGHTSPEED_STACK_IMAGE"
 

tests/test_document_processor_ogx.py

@@ -27,10 +27,9 @@
 image_name: starter
 
 apis:
-- agents
 - files
 - inference
-- safety
+- responses
 - tool_runtime
 - vector_io
 
@@ -50,21 +49,10 @@
       storage_dir: /tmp/files
     provider_id: meta-reference-files
     provider_type: inline::localfs
-  agents:
-  - config:
-      persistence:
-        agent_state:
-          namespace: agents_state
-          backend: kv_default
-        responses:
-          table_name: agents_responses
-          backend: sql_default
-    provider_id: meta-reference
-    provider_type: inline::meta-reference
   tool_runtime:
   - config: {{}}
-    provider_id: rag-runtime
-    provider_type: inline::rag-runtime
+    provider_id: file-search
+    provider_type: inline::file-search
   vector_io:
   - config:
       persistence:
@@ -107,8 +95,8 @@
   scoring_fns: []
   benchmarks: []
   tool_groups:
-  - toolgroup_id: builtin::rag
-    provider_id: rag-runtime
+  - toolgroup_id: builtin::file-search
+    provider_id: file-search
 """