chore: add pipeline mock development guide

This commit adds all the components needed to be able to mock the
pipelines.

dev/README.md

@@ -1,6 +1,6 @@
 # Development Guide
 
-This guide provides instructions for setting up the development environment and debugging the [AI worker](https://github.com/livepeer/ai-worker) repository.
+This document offers a comprehensive guide for configuring the development environment and debugging the [worker](worker) component within the [AI worker](https://github.com/livepeer/ai-worker) repository. For an in-depth exploration of developing the [runner](runner), please consult the [AI runner development guide](runner/dev/README.md).
 
 ## Debugging
 

runner/dev/README.md

@@ -85,3 +85,24 @@ To debug the AI runner when it operates within a container orchestrated by exter
    ```bash
    docker build -t livepeer/ai-runner:latest .
    ```
+
+### Mocking the Pipelines
+
+Mocking the pipelines is a practical approach for accelerating development and testing phases. This method simulates the pipeline execution, eliminating the need to run the actual model on a dedicated GPU. Follow the steps below to implement mocking:
+
+1. **Navigate to the Correct Directory**:
+   Ensure you are within the `runner` directory to apply changes effectively.
+
+2. **Applying the Mock Patch**:
+   Use the command below to apply the necessary code modifications for mocking the pipelines. This step introduces a mock environment for your development process.
+
+   ```bash
+   cd .. && git apply ./runner/dev/patches/mock.patch && cd runner
+   ```
+
+3. **Starting the AI Runner with Mocking**: Launch the AI runner with the environment variable `MOCK_PIPELINE` set to `True`. This enables the mock mode for pipeline execution.
+4. **Reverting Mock Changes**: Once testing is complete and you wish to return to the actual pipeline execution, revert the applied mock changes using the following command:
+
+   ```bash
+   cd .. && git apply -R ./runner/dev/patches/mock.patch && cd runner
+   ```

runner/dev/patches/mock.patch

@@ -0,0 +1,35 @@
+commit bd77a8bc1ec9afbc86682beff5b71ef13703753c
+Author: Rick Staa <[email protected]>
+Date:   Wed Jun 26 12:25:37 2024 +0100
+
+    refactor: improve code formating
+
+    This commit improves the I2I code formatting.
+
+diff --git a/runner/app/pipelines/text_to_image.py b/runner/app/pipelines/text_to_image.py
+index ac2334a..278c04e 100644
+--- a/runner/app/pipelines/text_to_image.py
++++ b/runner/app/pipelines/text_to_image.py
+@@ -14,7 +14,13 @@ from huggingface_hub import file_download, hf_hub_download
+ from safetensors.torch import load_file
+
+ from app.pipelines.base import Pipeline
+-from app.pipelines.util import get_model_dir, get_torch_device, SafetyChecker, is_lightning_model, is_turbo_model
++from app.pipelines.util import (
++    get_model_dir,
++    get_torch_device,
++    SafetyChecker,
++    is_lightning_model,
++    is_turbo_model,
++)
+
+ logger = logging.getLogger(__name__)
+
+@@ -52,7 +58,6 @@ class TextToImagePipeline(Pipeline):
+             logger.info("TextToImagePipeline using bfloat16 precision for %s", model_id)
+             kwargs["torch_dtype"] = torch.bfloat16
+
+-
+         # Special case SDXL-Lightning because the unet for SDXL needs to be swapped
+         if SDXL_LIGHTNING_MODEL_ID in model_id:
+             base = "stabilityai/stable-diffusion-xl-base-1.0"