docs: update customization

lavague-ai · May 21, 2024 · 38ab2aa · 38ab2aa
1 parent 21983f0
commit 38ab2aa
Showing 1 changed file with 7 additions and 6 deletions.
diff --git a/docs/docs/get-started/customization.md b/docs/docs/get-started/customization.md
@@ -5,17 +5,18 @@ Our Web Agents have the following customizable elemnets: LLM, multi-modal LLM (`
 These are the attributes of a `Context` object, which can be passed to the Action Engine and World Model.
 
 !!! info "Modifiable elements"
-    - `llm`: The `LLM` used by the `Action Engine` to translate text instructions into automation code. You can set the `llm` to any `LlamaIndex LLM object`.
 
-    - `mm_llm`: The `multi-modal LLM` used by the `World Model` to generate the next instruction to be enacted by the Action Engine based on the current state of the web page. You can set the `mm_llm` argument to any `LlamaIndex multi-modal LLM object`.
+    - **llm**: The `LLM` used by the `Action Engine` to translate text instructions into automation code. You can set the `llm` to any `LlamaIndex LLM object`.
 
-    - `embedding`: The `embedding model` is used by the `retriever` to convert segments of the HTML page of the target website into vectors, capturing semantic meaning. You can set this to any `LlamaIndex Embedding object`. 
+    - **mm_llm**: The `multi-modal LLM` used by the `World Model` to generate the next instruction to be enacted by the Action Engine based on the current state of the web page. You can set the `mm_llm` argument to any `LlamaIndex multi-modal LLM object`.
 
-    - `retriever`: The `retriever` is used within the `Action Model` to retrieve the most relevant HTML source code of the webpage to be able to generate the automation code targetting HTML elements. This can be any `LlamaIndex.retriever`.
+    - **embedding**: The `embedding model` is used by the `retriever` to convert segments of the HTML page of the target website into vectors, capturing semantic meaning. You can set this to any `LlamaIndex Embedding object`. 
 
-    - `prompt_template`: The prompt template used by the `Action Engine` to query the `LLM` in order to generate automation code for an instruction. You can replace this with your own prompt template as a string.
+    - **retriever**: The `retriever` is used within the `Action Model` to retrieve the most relevant HTML source code of the webpage to be able to generate the automation code targetting HTML elements. This can be any `LlamaIndex.retriever`.
 
-    - `extractor`: The `cleaning function` applied to the automation code generated by the LLM in the `Action Engine`. You can replace this with your own custom method as a callable.
+    - **prompt_template**: The prompt template used by the `Action Engine` to query the `LLM` in order to generate automation code for an instruction. You can replace this with your own prompt template as a string.
+
+    - **extractor**: The `cleaning function` applied to the automation code generated by the LLM in the `Action Engine`. You can replace this with your own custom method as a callable.
 
 These elements are initialized in a `Context` object, which can optionally passed to both the `Action Engine` and `World Model` used by an Agent. If you don't pass them your own Context object, the default OpenaiContext will be used.