Fix OpenAI for OnPrem Test Framework (#30)

* bump versions, conditional API button * bump llama_index * add cohere * Fix OPEN_API_KEY issue on testing framework
oracle-samples · Oct 16, 2024 · 53f2ef1 · 53f2ef1
1 parent 009813b
commit 53f2ef1
Show file tree

Hide file tree

Showing 14 changed files with 123 additions and 54 deletions.
diff --git a/app/requirements.txt b/app/requirements.txt
@@ -11,23 +11,23 @@
 ### pip3 install -r app/requirements.txt
 
 ## Top-Level brings in the required dependencies, if adding modules, try to find the minimum required
-giskard==2.15.1
 bokeh==3.6.0
 evaluate==0.4.3
-faiss-cpu==1.8.0.post1
-IPython==8.27.0
-langchain-cohere==0.3.0
-langchain-community==0.3.1
+faiss-cpu==1.9.0
+giskard==2.15.2
+IPython==8.28.0
+langchain-cohere==0.3.1
+langchain-community==0.3.2
 langchain-huggingface==0.1.0
 langchain-ollama==0.2.0
-langchain-openai==0.2.1
-llama_index==0.11.14
+langchain-openai==0.2.2
+llama_index==0.11.18
 lxml==5.3.0
 matplotlib==3.9.2
 oci>=2.0.0
 oracledb>=2.0.0
 plotly==5.24.1
-streamlit==1.38.0
+streamlit==1.39.0
 umap-learn==0.5.6
 
 ## For Licensing Purposes; ensures no GPU modules are installed

diff --git a/app/src/content/test_framework.py b/app/src/content/test_framework.py
@@ -151,6 +151,7 @@ def main():
         }
     </style>
     """
+
     st.markdown(css, unsafe_allow_html=True)
     st.title("Test Framework")
     # Used to clear the uploader files
@@ -232,7 +233,7 @@ def main():
         if qa_gen_file:
             qa_gen_button_disabled = False
 
-        left, right = st.columns(2, vertical_alignment="bottom")
+        left, right = st.columns([0.15, 0.85])
         if left.button("Generate Q&A", type="primary", key="qa_generation", disabled=qa_gen_button_disabled):
             placeholder = st.empty()
             with placeholder:
@@ -255,7 +256,7 @@ def main():
             # Generate Q&A
             qa_file = os.path.join(state["temp_dir"], f"{file_name}_{str(qa_count)}_test_set.json")
             state.qa_file = qa_file
-            state.test_set = utilities.generate_qa(qa_file, kb, qa_count)
+            state.test_set = utilities.generate_qa(qa_file, kb, qa_count, model=qa_llm, client=llm_client)
             placeholder.empty()
             st.success("Q&A Generation Succeeded.", icon="✅")
         right.button("Reset", key="reset_test_framework", type="primary", on_click=reset_test_set)
@@ -274,7 +275,7 @@ def main():
         if len(test_set_file) > 0:
             test_set_button_disabled = False
 
-        left, right = st.columns(2, vertical_alignment="bottom")
+        left, right = st.columns([0.15, 0.85], vertical_alignment="bottom")
         if left.button("Load Tests", key="load_tests", disabled=test_set_button_disabled):
             placeholder = st.empty()
             with placeholder:

diff --git a/app/src/modules/chatbot_server.py b/app/src/modules/chatbot_server.py
@@ -204,9 +204,10 @@ def gui_start():
 # ChatBot Sidebar
 ###################################
 def chatbot_sidebar():
-    st.session_state["port"] = st.sidebar.number_input(
-        "Enter the port number for the chatbot server:", value=8000, min_value=1, max_value=65535
-    )
-    st.session_state["api_key"] = st.sidebar.text_input("API_KEY", type="password", value="abc")
-    st.sidebar.button("Start server", type="primary", on_click=gui_start)
-    st.sidebar.divider()
+    if not state.disable_api:
+        st.session_state["port"] = st.sidebar.number_input(
+            "Enter the port number for the chatbot server:", value=8000, min_value=1, max_value=65535
+        )
+        st.session_state["api_key"] = st.sidebar.text_input("API_KEY", type="password", value="abc")
+        st.sidebar.button("Start server", type="primary", on_click=gui_start)
+        st.sidebar.divider()
diff --git a/app/src/modules/logging_config.py b/app/src/modules/logging_config.py
@@ -25,7 +25,8 @@ def setup_logging():
     logging.getLogger("PIL").setLevel(logging.INFO)
     logging.getLogger("openai").setLevel(logging_level)
     logging.getLogger("httpcore").setLevel(logging_level)
-
+    # Sagemaker continuously complains about config, suppress
+    logging.getLogger("sagemaker.config").setLevel(logging.WARNING)
 
 # Call setup_logging when this module is imported
 setup_logging()
diff --git a/app/src/modules/st_common.py b/app/src/modules/st_common.py
@@ -515,6 +515,11 @@ def empty_key(obj):
     state_dict_filt = {key: state_dict[key] for key in include_keys if key in state_dict}
     state_dict_filt = empty_key(state_dict_filt)
     session_state_json = json.dumps(state_dict_filt, indent=4)
-    st.sidebar.download_button(
-        label="Download Settings", data=session_state_json, file_name="sandbox_settings.json", use_container_width=True
-    )
+    # Only allow exporting settings if tools/admin is enabled
+    if not state.disable_tools and not state.disable_admin:
+        st.sidebar.download_button(
+            label="Download Settings",
+            data=session_state_json,
+            file_name="sandbox_settings.json",
+            use_container_width=True,
+        )
diff --git a/app/src/modules/utilities.py b/app/src/modules/utilities.py
@@ -287,7 +287,7 @@ def populate_vs(
     model_name,
     distance_metric,
     input_data: Union[List["LangchainDocument"], List] = None,
-    rate_limit = 0
+    rate_limit=0,
 ):
     """Populate the Vector Storage"""
 
@@ -350,7 +350,7 @@ def json_to_doc(file: str):
             "Processing: %i Chunks of %i (Rate Limit: %i)",
             len(unique_chunks) if len(unique_chunks) < i + batch_size else i + batch_size,
             len(unique_chunks),
-            rate_limit
+            rate_limit,
         )
         OracleVS.add_documents(vectorstore, documents=batch)
         if rate_limit > 0:
@@ -700,11 +700,11 @@ def build_knowledge_base(text_nodes, kb_file, llm_client, embed_client):
     return knowledge_base
 
 
-def generate_qa(qa_file, kb, qa_count, api="openai", model="gpt-4o-mini"):
+def generate_qa(qa_file, kb, qa_count, api="openai", model="gpt-4o-mini", client=None):
     """Generate an example QA"""
-    logger.info("QA Generation starting..")
+    logger.info("QA Generation starting.. (model=%s, client=%s)", model, client)
     set_llm_api(api)
-    set_default_client(OpenAIClient(model=model))
+    set_default_client(OpenAIClient(model=model, client=client))
 
     test_set = generate_testset(
         kb,

diff --git a/app/src/oaim-sandbox.py b/app/src/oaim-sandbox.py
@@ -26,11 +26,13 @@
 os.environ["GSK_DISABLE_SENTRY"] = "true"
 os.environ["GSK_DISABLE_ANALYTICS"] = "true"
 
+
 #############################################################################
 # MAIN
 #############################################################################
 def main():
     """Streamlit GUI"""
+    st.set_page_config(layout="wide")
     # initialize Components
     db_initialize()
     model_initialize()
@@ -48,7 +50,6 @@ def main():
     # GUI Defaults
     css = """
     <style>
-        section.main > div {max-width:65rem; padding-top: 3.85rem;}
         section[data-testid="stSidebar"] div.stButton button {
             width: 100%;
         }
@@ -61,12 +62,14 @@ def main():
     st.markdown(css, unsafe_allow_html=True)
     st.logo("images/logo_light.png")
 
-    # Page Definition
+    # Enable/Disable Functionality
+    state.disable_api = os.environ.get("DISABLE_API", "false").lower() == "true"
     state.disable_tools = os.environ.get("DISABLE_TOOLS", "false").lower() == "true"
     state.disable_tests = os.environ.get("DISABLE_TESTS", "false").lower() == "true" and not state.disable_tools
     state.disable_admin = os.environ.get("DISABLE_ADMIN", "false").lower() == "true" and not state.disable_tools
     state.disable_oci = os.environ.get("DISABLE_OCI", "false").lower() == "true" and not state.disable_admin
 
+    # Left Hand Side - Navigation
     chatbot = st.Page("content/chatbot.py", title="ChatBot", icon="💬", default=True)
     navigation = {
         "": [chatbot],
@@ -81,22 +84,22 @@ def main():
         navigation["Tools"] = [prompt_eng]
 
     # Administration
-    import_settings = st.Page("content/import_settings.py", title="Import Settings", icon="💾")
-    navigation["Configuration"] = [import_settings]
     if not state.disable_tools and not state.disable_admin:
         # Define Additional Pages
+        import_settings = st.Page("content/import_settings.py", title="Import Settings", icon="💾")
         split_embed = st.Page("content/split_embed.py", title="Split/Embed", icon="📚")
         model_config = st.Page("content/model_config.py", title="Models", icon="🤖")
         db_config = st.Page("content/db_config.py", title="Database", icon="🗄️")
         # Update Navigation
         navigation["Tools"].insert(0, split_embed)
-        navigation["Configuration"].insert(0, model_config)
+        navigation["Configuration"] = [model_config]
         navigation["Configuration"].insert(1, db_config)
+        navigation["Configuration"].insert(2, import_settings)
         if not state.disable_oci:
             oci_config = st.Page("content/oci_config.py", title="OCI", icon="☁️")
             navigation["Configuration"].insert(2, oci_config)
 
-    pg = st.navigation(navigation)
+    pg = st.navigation(navigation, position="sidebar", expanded=False)
     pg.run()
 
 

diff --git a/docs/content/_index.md b/docs/content/_index.md
@@ -9,6 +9,8 @@ keywords = 'oracle microservices development genai rag'
 <!--
 Copyright (c) 2023, 2024, Oracle and/or its affiliates.
 Licensed under the Universal Permissive License v1.0 as shown at http://oss.oracle.com/licenses/upl.
+
+spell-checker:ignore streamlit, genai, venv, oaim
 -->
 
 {{% notice style="code" title="10-Sept-2024: Documentation In-Progress..." icon="pen" %}}

diff --git a/docs/content/sandbox/configuration/model_config.md b/docs/content/sandbox/configuration/model_config.md
@@ -16,21 +16,25 @@ At a minimum, a Large _Language Model_ (LLM) must be configured in **Oracle AI M
 If there is a specific model that you would like to use with the **Oracle AI Microservices Sandbox**, please [open an issue in GitHub](https://github.com/oracle-samples/oaim-sandbox/issues/new).
 {{% /notice %}}
 
-| Model                          | Type  | API                                                      | On-Premises |
-| ------------------------------ | ----- | -------------------------------------------------------- | ----------- |
-| llama3.1                       | LLM   | [ChatOllama](#additional-information)                    | X           |
-| gpt-3.5-turbo                  | LLM   | [OpenAI](#additional-information)                        |             |
-| gpt-4o-mini                    | LLM   | [OpenAI](#additional-information)                        |             |
-| gpt-4                          | LLM   | [OpenAI](#additional-information)                        |             |
-| gpt-4o                         | LLM   | [OpenAI](#additional-information)                        |             |
-| llama-3-sonar-small-32k-chat   | LLM   | [ChatPerplexity](#additional-information)                |             |
-| llama-3-sonar-small-32k-online | LLM   | [ChatPerplexity](#additional-information)                |             |
-| mxbai-embed-large              | Embed | [OllamaEmbeddings](#additional-information)              | X           |
-| nomic-embed-text               | Embed | [OllamaEmbeddings](#additional-information)              | X           |
-| all-minilm                     | Embed | [OllamaEmbeddings](#additional-information)              | X           |
-| thenlper/gte-base              | Embed | [HuggingFaceEndpointEmbeddings](#additional-information) | X           |
-| text-embedding-3-small         | Embed | [OpenAIEmbeddings](#additional-information)              |             |
-| text-embedding-3-large         | Embed | [OpenAIEmbeddings](#additional-information)              |             |
+| Model                           | Type  | API                                                      | On-Premises |
+| ------------------------------- | ----- | -------------------------------------------------------- | ----------- |
+| llama3.1                        | LLM   | [ChatOllama](#additional-information)                    | X           |
+| gpt-3.5-turbo                   | LLM   | [OpenAI](#additional-information)                        |             |
+| gpt-4o-mini                     | LLM   | [OpenAI](#additional-information)                        |             |
+| gpt-4                           | LLM   | [OpenAI](#additional-information)                        |             |
+| gpt-4o                          | LLM   | [OpenAI](#additional-information)                        |             |
+| llama-3-sonar-small-128k-chat   | LLM   | [ChatPerplexity](#additional-information)                |             |
+| llama-3-sonar-small-128k-online | LLM   | [ChatPerplexity](#additional-information)                |             |
+| command-r                       | LLM   | [Cohere](#additional-information)                        |             |
+| mxbai-embed-large               | Embed | [OllamaEmbeddings](#additional-information)              |             |
+| nomic-embed-text                | Embed | [OllamaEmbeddings](#additional-information)              | X           |
+| all-minilm                      | Embed | [OllamaEmbeddings](#additional-information)              | X           |
+| thenlper/gte-base               | Embed | [HuggingFaceEndpointEmbeddings](#additional-information) | X           |
+| text-embedding-3-small          | Embed | [OpenAIEmbeddings](#additional-information)              |             |
+| text-embedding-3-large          | Embed | [OpenAIEmbeddings](#additional-information)              |             |
+| embed-english-v3.0              | Embed | [CohereEmbeddings](#additional-information)              |             |
+| embed-english-light-v3.0        | Embed | [CohereEmbeddings](#additional-information)              |             |
+
 
 ## Configuration
 
@@ -148,6 +152,20 @@ Example of running thenlper/gte-base in a container:
 
    **NOTE:** if there is no IP, use 127.0.0.1
 {{% /tab %}}
+{{% tab title="Cohere" %}}
+# Cohere
+
+[Cohere](https://cohere.com/) is an AI-powered answer engine. To use Cohere, you will need to sign-up and provide the **Sandbox** an API Key.  Cohere offers a free-trial, rate-limited API Key.
+
+**WARNING:** Cohere is a cloud model and you should familiarize yourself with their Privacy Policies if using it to experiment with private, sensitive data in the **Sandbox**.
+
+>[!code]Skip the GUI!
+>You can set the following environment variable to automatically set the `API Key` and enable Perplexity models:
+:
+>```shell
+>export COHERE_API_KEY=<super-secret API Key>
+>```
+{{% /tab %}}
 {{% tab title="OpenAI" %}}
 # OpenAI
 

diff --git a/helm/templates/_helpers.tpl b/helm/templates/_helpers.tpl
@@ -78,4 +78,26 @@ Set the path based on baseUrlPath
 {{- else -}}
 {{- $baseUrlPath -}}
 {{- end -}}
+{{- end -}}
+
+{{/*
+Set default value for mtlsWallet if not defined
+*/}}
+{{- define "getMtlsWallet" -}}
+{{- if .Values.oaimSandbox.database.mtlsWallet -}}
+  {{- .Values.oaimSandbox.database.mtlsWallet -}}
+{{- else -}}
+  {{- dict -}}  # Return an empty dictionary if mtlsWallet is not defined
+{{- end -}}
+{{- end -}}
+
+{{/*
+Set default value for tnsAdmin if not defined
+*/}}
+{{- define "getTnsAdmin" -}}
+{{- if .Values.oaimSandbox.database.tnsAdmin -}}
+  {{- .Values.oaimSandbox.database.tnsAdmin -}}
+{{- else -}}
+  {{- dict -}}  # Return an empty dictionary if tnsAdmin is not defined
+{{- end -}}
 {{- end -}}
diff --git a/helm/templates/adb.yaml b/helm/templates/adb.yaml
@@ -4,7 +4,6 @@
 {{- if .Values.oaimSandbox.database.ocid }}
 apiVersion: database.oracle.com/v1alpha1
 kind: AutonomousDatabase
-metadata:
 metadata:
   name: {{ .Values.oaimSandbox.database.name | lower }}-adb
   labels:

diff --git a/helm/templates/configmap.yaml b/helm/templates/configmap.yaml
@@ -21,7 +21,10 @@ data:
     cookieSecret = "{{ .Chart.Name }}-cookie"
     headless = true
     fileWatcherType = "none"
+    {{- $path := include "getPath" . }}
+    {{- if ne $path "/" }}
     baseUrlPath = {{ include "getPath" . | quote }}
+    {{- end }}
 
     [client]
     toolbarMode = "minimal"

diff --git a/helm/templates/deployment.yaml b/helm/templates/deployment.yaml
@@ -86,12 +86,13 @@ spec:
                 secretKeyRef:
                   name: {{ .Values.oaimSandbox.database.authN.secretName | quote }}
                   key: {{ .Values.oaimSandbox.database.authN.serviceKey | quote }}
-            {{- if .Values.oaimSandbox.database.mtlsWallet.secretName }}
+            {{- $mtlsWallet := include "getMtlsWallet" . | fromJson }}
+            {{- if $mtlsWallet.secretName }}
             - name: DB_WALLET_PASSWORD
               valueFrom:
                 secretKeyRef:
-                  name: {{ .Values.oaimSandbox.database.mtlsWallet.secretName | quote }}
-                  key: {{ .Values.oaimSandbox.database.mtlsWallet.secretKey | quote }}
+                  name: {{ $mtlsWallet.secretName | quote }}
+                  key: {{ $mtlsWallet.secretKey | quote }}
             {{ else }}
             - name: DB_WALLET_PASSWORD
               valueFrom:
@@ -113,6 +114,10 @@ spec:
             - name: DISABLE_ADMIN
               value: "true"
             {{- end }}
+            {{- if eq .disableAAPI "true" }}
+            - name: DISABLE_API
+              value: "true"
+            {{- end }}
             {{- if .models }}
             {{- if .models.openAI.secretName }}
             - name: OPENAI_API_KEY
@@ -128,6 +133,13 @@ spec:
                   name: {{ .models.perplexity.secretName }}
                   key: {{ .models.perplexity.secretKey }}
             {{- end }}
+            {{- if .models.cohere.secretName }}
+            - name: COHERE_API_KEY
+              valueFrom:
+                secretKeyRef:
+                  name: {{ .models.cohere.secretName }}
+                  key: {{ .models.cohere.secretKey }}
+            {{- end }}
             {{- end }}
             {{- end }}
 
@@ -156,8 +168,9 @@ spec:
         {{- if .Values.oaimSandbox.database.enabled }}
         - name: tns-admin
           secret:
-            {{- if .Values.oaimSandbox.database.tnsAdmin.secretName }}
-            secretName: {{ .Values.oaimSandbox.database.tnsAdmin.secretName | quote }}
+            {{- $tnsAdmin := include "getTnsAdmin" . | fromJson }}
+            {{- if $tnsAdmin.secretName }}
+            secretName: {{ $tnsAdmin.secretName | quote }}
             {{ else }}
             secretName: {{ .Values.oaimSandbox.database.name | lower }}-tns-admin-{{ .Release.Revision }}
             {{- end }}

diff --git a/helm/templates/secret.yaml b/helm/templates/secret.yaml
@@ -1,7 +1,8 @@
 ## Copyright (c) 2023, 2024, Oracle and/or its affiliates.
 ## Licensed under the Universal Permissive License v1.0 as shown at http://oss.oracle.com/licenses/upl.
 
-{{- if not .Values.oaimSandbox.database.mtlsWallet.secretName }}
+{{- $mtlsWallet := include "getMtlsWallet" . | fromJson }}
+{{- if not $mtlsWallet.secretName }}
 apiVersion: v1
 kind: Secret
 metadata: