fix: config allows overriding LLM

examples/high_level/highlevel.py

@@ -93,6 +93,27 @@ def custom_openai_llm():
             continue
         print(client.query(prompt, include_own=True))
 
+def custom_llama_llm():
+    """
+    You're not restricted to any LLM provider, you are free to use any of the supported BaseChatModel 
+    (https://python.langchain.com/docs/integrations/chat/)
+
+    In this example we're using groq to run Llama 3.1 70b, but this can also be ran on your own hardware
+    for a totally private RAG!
+    """
+    # pip install langchain-groq
+    from langchain_groq import ChatGroq
+
+    llm = ChatGroq(model="llama3-groq-70b-8192-tool-use-preview")
+    config = NyxConfigExtended.from_env(provider=ConfigType.BASE)
+
+    client = NyxLangChain(config=config, llm=llm)
+    while True:
+        prompt = input("What is your question? ")
+        if prompt == "":
+            continue
+        print(client.query(prompt, include_own=True))
+
 
 if __name__ == "__main__":
     custom_openai_llm()

nyx_client/nyx_client/configuration.py

@@ -142,7 +142,5 @@ def from_env(
                 api_key = os.environ["OPENAI_API_KEY"]
             case ConfigType.COHERE:
                 api_key = os.environ["COHERE_API_KEY"]
-            case other:
-                raise Exception(f"{other} is not a config type")
         base = BaseNyxConfig.from_env(env_file, override_token)
         return NyxConfigExtended(api_key=api_key, provider=provider, base_config=base)

nyx_extras/nyx_extras/utils.py

@@ -14,6 +14,7 @@
 
 """Module that contains utility functions, as well as tooling for manual parsing of data contained in Nyx."""
 
+import json
 import logging
 import os
 import sqlite3
@@ -191,6 +192,8 @@ def data_as_db(
                     log.warning("%s is unsupported type %s", d.title, d.content_type)
                     continue
                 content.columns = Parser.normalise_values(content.columns)
+                for col in content.select_dtypes(include=["object"]).columns:
+                    content[col] = content[col].apply(lambda x: json.dumps(x) if isinstance(x, (list, dict)) else x)
                 content.to_sql(table_name, db_engine)
             except pd.errors.ParserError:
                 if d.content_type in ["csv", "json", *Parser._excel_mimes]: