Sync with downstream

ols/src/llms/providers/azure_openai.py

@@ -74,6 +74,7 @@ def default_params(self) -> dict[str, Any]:
             "temperature": 0.01,
             "max_tokens": 512,
             "verbose": False,
+            "http_client": self._construct_httpx_client(False),
         }
 
         if self.credentials is not None:

ols/src/llms/providers/openai.py

@@ -46,6 +46,7 @@ def default_params(self) -> dict[str, Any]:
             "temperature": 0.01,
             "max_tokens": 512,
             "verbose": False,
+            "http_client": self._construct_httpx_client(False),
         }
 
     def load(self) -> LLM:

ols/src/llms/providers/provider.py

@@ -50,6 +50,7 @@ class ProviderParameter:
     ProviderParameter("temperature", float),
     ProviderParameter("max_tokens", int),
     ProviderParameter("verbose", bool),
+    ProviderParameter("http_client", httpx.Client),
 }
 
 OpenAIParameters = {
@@ -66,6 +67,7 @@ class ProviderParameter:
     ProviderParameter("temperature", float),
     ProviderParameter("max_tokens", int),
     ProviderParameter("verbose", bool),
+    ProviderParameter("http_client", httpx.Client),
 }
 
 RHOAIVLLMParameters = {
@@ -317,3 +319,11 @@ def _override_params(self, params: dict[Any, Any]) -> dict[Any, Any]:
             updated_params = {**updated_params, **config.dev_config.llm_params}
 
         return updated_params
+
+    def _construct_httpx_client(
+        self, use_custom_certificate_store: bool
+    ) -> httpx.Client:
+        """Construct HTTPX client instance to be used to communicate with LLM."""
+        if use_custom_certificate_store:
+            return httpx.Client(verify=self.provider_config.certificates_store)
+        return httpx.Client()

ols/src/llms/providers/rhelai_vllm.py

@@ -3,7 +3,6 @@
 import logging
 from typing import Any, Optional
 
-import httpx
 from langchain.llms.base import LLM
 from langchain_openai import ChatOpenAI
 
@@ -48,7 +47,7 @@ def default_params(self) -> dict[str, Any]:
             "temperature": 0.01,
             "max_tokens": 512,
             "verbose": False,
-            "http_client": httpx.Client(verify=self.provider_config.certificates_store),
+            "http_client": self._construct_httpx_client(True),
         }
 
     def load(self) -> LLM:

ols/src/llms/providers/rhoai_vllm.py

@@ -3,7 +3,6 @@
 import logging
 from typing import Any, Optional
 
-import httpx
 from langchain.llms.base import LLM
 from langchain_openai import ChatOpenAI
 
@@ -48,7 +47,7 @@ def default_params(self) -> dict[str, Any]:
             "temperature": 0.01,
             "max_tokens": 512,
             "verbose": False,
-            "http_client": httpx.Client(verify=self.provider_config.certificates_store),
+            "http_client": self._construct_httpx_client(True),
         }
 
     def load(self) -> LLM:

ols/src/prompts/prompts.py

@@ -5,13 +5,6 @@
 
 from ols.constants import SUBJECT_ALLOWED, SUBJECT_REJECTED
 
-# Note::
-# Right now templates are somewhat alligned to make granite work better.
-# GPT still works well with this. Ideally we should have model specific tags.
-# For history we can laverage ChatPromptTemplate from langchain,
-# but that is not done as granite was adding role tags like `Human:` in the response.
-# With PromptTemplate, we have more control how we want to structure the prompt.
-
 QUERY_SYSTEM_INSTRUCTION = """
 You are OpenShift Lightspeed - an intelligent assistant for question-answering tasks \
 related to the OpenShift container orchestration platform.

scripts/build-container.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 # Version
-OLS_VERSION=v0.0.0
+OLS_VERSION=v0.2.1
 
 # To build container for local use
 if [ -z "$OLS_NO_IMAGE_CACHE" ]; then

tests/config/operator_install/imagedigestmirrorset.yaml

@@ -5,15 +5,15 @@ metadata:
   name: openshift-lightspeed-prod-to-ci
 spec:
   imageDigestMirrors:
-    - source: registry.redhat.io/openshift-lightspeed-beta/lightspeed-rhel9-operator
+    - source: registry.redhat.io/openshift-lightspeed-tech-preview/lightspeed-rhel9-operator
       mirrors:
         - quay.io/redhat-user-workloads/crt-nshift-lightspeed-tenant/ols/lightspeed-operator
-    - source: registry.redhat.io/openshift-lightspeed-beta/lightspeed-operator-bundle
+    - source: registry.redhat.io/openshift-lightspeed-tech-preview/lightspeed-operator-bundle
       mirrors:
         - quay.io/redhat-user-workloads/crt-nshift-lightspeed-tenant/ols/bundle
-    - source: registry.redhat.io/openshift-lightspeed-beta/lightspeed-service-api-rhel9
+    - source: registry.redhat.io/openshift-lightspeed-tech-preview/lightspeed-service-api-rhel9
       mirrors:
         - quay.io/redhat-user-workloads/crt-nshift-lightspeed-tenant/ols/lightspeed-service
-    - source: registry.redhat.io/openshift-lightspeed-beta/lightspeed-console-plugin-rhel9
+    - source: registry.redhat.io/openshift-lightspeed-tech-preview/lightspeed-console-plugin-rhel9
       mirrors:
         - quay.io/redhat-user-workloads/crt-nshift-lightspeed-tenant/ols/lightspeed-console

tests/unit/llms/providers/test_azure_openai.py

@@ -3,6 +3,7 @@
 import time
 from unittest.mock import patch
 
+import httpx
 import pytest
 from azure.core.credentials import AccessToken
 from langchain_openai import AzureChatOpenAI
@@ -211,6 +212,13 @@ def test_basic_interface(provider_config):
 
     assert azure_openai.default_params["azure_endpoint"] == "test_url"
 
+    # check the HTTP client parameter
+    assert "http_client" in azure_openai.default_params
+    assert azure_openai.default_params["http_client"] is not None
+
+    client = azure_openai.default_params["http_client"]
+    assert isinstance(client, httpx.Client)
+
 
 def test_credentials_in_directory_handling(provider_config_credentials_directory):
     """Test that credentials in directory is handled as expected."""

tests/unit/llms/providers/test_openai.py

@@ -1,5 +1,6 @@
 """Unit tests for OpenAI provider."""
 
+import httpx
 import pytest
 from langchain_openai.chat_models.base import ChatOpenAI
 
@@ -81,6 +82,13 @@ def test_basic_interface(provider_config):
     assert "model" in openai.default_params
     assert "max_tokens" in openai.default_params
 
+    # check the HTTP client parameter
+    assert "http_client" in openai.default_params
+    assert openai.default_params["http_client"] is not None
+
+    client = openai.default_params["http_client"]
+    assert isinstance(client, httpx.Client)
+
 
 def test_params_handling(provider_config):
     """Test that not allowed parameters are removed before model init."""

tests/unit/llms/providers/test_rhelai_vllm.py

@@ -2,6 +2,7 @@
 
 import os
 
+import httpx
 import pytest
 from langchain_openai.chat_models.base import ChatOpenAI
 
@@ -99,9 +100,14 @@ def test_basic_interface(provider_config, fake_certifi_store):
     assert "base_url" in rhelai_vllm.default_params
     assert "model" in rhelai_vllm.default_params
     assert "max_tokens" in rhelai_vllm.default_params
+
+    # check the HTTP client parameter
     assert "http_client" in rhelai_vllm.default_params
     assert rhelai_vllm.default_params["http_client"] is not None
 
+    client = rhelai_vllm.default_params["http_client"]
+    assert isinstance(client, httpx.Client)
+
 
 def test_params_handling(provider_config, fake_certifi_store):
     """Test that not allowed parameters are removed before model init."""

tests/unit/llms/providers/test_rhoai_vllm.py

@@ -2,6 +2,7 @@
 
 import os
 
+import httpx
 import pytest
 from langchain_openai.chat_models.base import ChatOpenAI
 
@@ -99,9 +100,14 @@ def test_basic_interface(provider_config, fake_certifi_store):
     assert "base_url" in rhoai_vllm.default_params
     assert "model" in rhoai_vllm.default_params
     assert "max_tokens" in rhoai_vllm.default_params
+
+    # check the HTTP client parameter
     assert "http_client" in rhoai_vllm.default_params
     assert rhoai_vllm.default_params["http_client"] is not None
 
+    client = rhoai_vllm.default_params["http_client"]
+    assert isinstance(client, httpx.Client)
+
 
 def test_params_handling(provider_config, fake_certifi_store):
     """Test that not allowed parameters are removed before model init."""