Added instructions for configuring Ollama models (#38)

* Update local-models.md

* Update docs/advanced-usage/local-models.md

Co-authored-by: ellipsis-dev[bot] <65095814+ellipsis-dev[bot]@users.noreply.github.com>

---------

Co-authored-by: Matt Rubens <[email protected]>
Co-authored-by: ellipsis-dev[bot] <65095814+ellipsis-dev[bot]@users.noreply.github.com>

Author: 3 people

docs/advanced-usage/local-models.md

@@ -23,39 +23,62 @@ Roo Code currently supports two main local model providers:
 
 ## Setting Up Ollama
 
-1.  **Download and Install Ollama:**  Download the Ollama installer for your operating system from the [Ollama website](https://ollama.com/). Follow the installation instructions.
+1.  **Download and Install Ollama:**  Download the Ollama installer for your operating system from the [Ollama website](https://ollama.com/). Follow the installation instructions. Make sure Ollama is running
+
+    ```bash
+    ollama serve
+    ```
 
 2.  **Download a Model:**  Ollama supports many different models.  You can find a list of available models on the [Ollama website](https://ollama.com/library).  Some recommended models for coding tasks include:
 
     *   `codellama:7b-code` (good starting point, smaller)
     *   `codellama:13b-code` (better quality, larger)
     *   `codellama:34b-code` (even better quality, very large)
+    *   `qwen2.5-coder:32b`
     *   `mistralai/Mistral-7B-Instruct-v0.1` (good general-purpose model)
     *   `deepseek-coder:6.7b-base` (good for coding tasks)
     * `llama3:8b-instruct-q5_1` (good for general tasks)
 
     To download a model, open your terminal and run:
 
     ```bash
-    ollama run <model_name>
+    ollama pull <model_name>
     ```
 
     For example:
 
     ```bash
-    ollama run codellama:7b-code
+    ollama pull qwen2.5-coder:32b
+    ```
+
+3. **Configure the Model:** by default, Ollama uses a context window size of 2048 tokens, which is too small for Roo Code requests. You need to have at least 12k to get decent results, ideally - 32k. To configure a model, you actually need to set its parameters and save a copy of it.
+
+   Load the model (we will use `qwen2.5-coder:32b` as an example):
+   
+    ```bash
+    ollama run qwen2.5-coder:32b
+    ```
+
+   Change context size parameter:
+
+    ```bash
+    /set parameter num_ctx 32768
+    ```
+
+    Save the model with a new name:
+
+    ```bash
+    /save your_model_name
     ```
-    
-    This will download the model and then run it in your terminal.  You can type `/bye` to exit the model.
-    
-    **Important:** The first time you run a model, Ollama *must* be running. If you run with a model and Ollama is not open, VSCode may hang.
+      
 
-3.  **Configure Roo Code:**
+4.  **Configure Roo Code:**
     *   Open the Roo Code sidebar (🚀 icon).
     *   Click the settings gear icon (⚙️).
     *   Select "ollama" as the API Provider.
-    *   Enter the Model ID (e.g., `codellama:7b-code`).
+    *   Enter the Model name from the previous step (e.g., `your_model_name`).
     *   (Optional) You can configure the base URL if you're running Ollama on a different machine. The default is `http://localhost:11434`.
+    *   (Optional) Configure Model context size in Advanced settings, so Roo Code knows how to manage its sliding window.
 
 ## Setting Up LM Studio
 
@@ -85,4 +108,4 @@ Roo Code currently supports two main local model providers:
 
 *   **Slow Response Times:** Local models can be slower than cloud-based models, especially on less powerful hardware.  If performance is an issue, try using a smaller model.
 
-*   **Model Not Found:** Ensure you have typed in the name of the model correctly. If you're using Ollama, use the same name that you provide in the `ollama run` command.
+*   **Model Not Found:** Ensure you have typed in the name of the model correctly. If you're using Ollama, use the same name that you provide in the `ollama run` command.