From 8d5da7a5cd111500efecf5de8aa4c41f0b4cb3fd Mon Sep 17 00:00:00 2001
From: lasseedfast <lasse.edfast@gmail.com>
Date: Mon, 7 Oct 2024 17:24:40 +0200
Subject: [PATCH] Refactor LLM class initialization and add note about default
 value of num_ctx parameter

---
 README.md        |  4 +++-
 highlight_pdf.py | 11 ++++-------
 2 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/README.md b/README.md
index 9ef0161..9846843 100644
--- a/README.md
+++ b/README.md
@@ -167,11 +167,13 @@ A Streamlit example is provided in `example_streamlit_app.py` to demonstrate how
 
 #### Methods
 
-- `__init__(self, openai_key=False, model=None, temperature=0, system_prompt=None, num_ctx=None, memory=True, keep_alive=3600)`: Initializes the LLM class with the provided parameters.
+- `__init__(self, openai_key=False, model=None, temperature=0, system_prompt=None, num_ctx=None, memory=True, keep_alive=3600)`: Initializes the LLM class with the provided parameters. 
 - `use_openai(self, key, model)`: Configures the class to use OpenAI for generating responses.
 - `use_ollama(self, model)`: Configures the class to use Ollama for generating responses.
 - `async generate(self, prompt)`: Asynchronously generates a response based on the provided prompt.
 
+**Note:** The `num_ctx` parameter is set to 20000 by default, which may not be sufficient for all use cases. Adjust this value based on your specific requirements.
+
 ## Contributing
 
 Contributions are welcome! Please open an issue or submit a pull request for any improvements or bug fixes.
diff --git a/highlight_pdf.py b/highlight_pdf.py
index 533550a..ea9dfaf 100644
--- a/highlight_pdf.py
+++ b/highlight_pdf.py
@@ -46,8 +46,8 @@ class LLM:
 
     Attributes:
         model (str): The model to be used for generating responses.
+        num_ctx (int): The number of context tokens to be used. Defaults to 20000.
         temperature (float): The temperature setting for the model's response generation.
-        num_ctx (int): The number of context tokens to be used.
         keep_alive (int): The keep-alive duration for the connection.
         options (dict): Options for the model's response generation.
         memory (bool): Whether to retain conversation history.
@@ -70,11 +70,11 @@ class LLM:
 
     def __init__(
         self,
+        num_ctx=20000,
         openai_key=False,
         model=None,
         temperature=0,
         system_prompt=None,
-        num_ctx=None,
         memory=True,
         keep_alive=3600,
     ):
@@ -86,7 +86,7 @@ class LLM:
         model (str, optional): The model to be used. Defaults to None.
         temperature (float, optional): Sampling temperature for the model. Defaults to 0.
         system_prompt (str, optional): Initial system prompt for the model. Defaults to None.
-        num_ctx (int, optional): Number of context tokens. Defaults to None.
+        context_window (int, optional): Number of context tokens. Defaults to None.
         memory (bool, optional): Whether to use memory. Defaults to True.
         keep_alive (int, optional): Keep-alive duration in seconds. Defaults to 3600.
         """
@@ -96,12 +96,9 @@ class LLM:
         else:
             self.model = os.getenv("LLM_MODEL")
         self.temperature = temperature
-        self.num_ctx = num_ctx
         self.keep_alive = keep_alive
-        self.options = {"temperature": self.temperature}
+        self.options = {"temperature": self.temperature, num_ctx: num_ctx}
         self.memory = memory
-        if self.num_ctx:
-            self.options["num_ctx"] = self.num_ctx
         if system_prompt:
             self.messages = [{"role": "system", "content": system_prompt}]
         else: