Tweaked to remove the deterministic thread as not relevant to my project, and added safety steps for the baseline model.

config/architectures.json

@@ -4,14 +4,9 @@
       "name": "Baseline LLM",
       "description": "Just a direct call through to the LLM without any additional components.",
       "steps": [
-        {"class":  "HFLlamaHttpRequestor", "params":  {"model":  "meta-llama/Llama-2-7b-chat-hf", "system_prompt":  "You are a helpful agent.", "max_tokens":  2000}}
-      ]
-    },
-    {
-      "name": "Determinstic LLM",
-      "description": "This is just a demonstration setup for configuration of the temperature setting.  In this architecture setup, the temperature has been set to 0.001 which means the LLM component is in practical terms, not selecting probabilistically, rather deterministically.  Therefore the same request should always result in the same response. In order to see this, copy your query, try it, then navigate away and back before trying it again - you should see the same response.",
-      "steps": [
-        {"class":  "HFLlamaHttpRequestor", "params":  {"model":  "meta-llama/Llama-2-7b-chat-hf", "system_prompt":  "You are a helpful agent.", "max_tokens":  2000, "temperature":  0.01}}
+        {"class":  "InputRequestScreener"},
+        {"class":  "HFLlamaHttpRequestor", "params":  {"model":  "meta-llama/Llama-2-7b-chat-hf", "system_prompt":  "You are a helpful agent.", "max_tokens":  2000}},
+        {"class":  "OutputResponseScreener"}
       ]
     },
     {