### Quick Start Installation

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/README.md

Installs the openpilot software using a curl command. This is the primary method for getting started with openpilot on supported hardware.

```bash
bash <(curl -fsSL openpilot.comma.ai)
```

--------------------------------

### Run openpilot Setup Script

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tools/README.md

Navigates into the cloned openpilot directory and executes the primary setup script. This script handles initial environment configuration and dependency installation.

```bash
cd openpilot
tools/op.sh setup
```

--------------------------------

### Setup Openpilot Development Environment

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/docs/how-to/turn-the-speed-blue.md

Clones the openpilot repository and installs all necessary dependencies using a provided bash script. This is the initial step for setting up a local development environment.

```bash
bash <(curl -fsSL openpilot.comma.ai)
```

--------------------------------

### Import necessary libraries

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/docs/quickstart.md

Imports required modules for numerical operations and timing, setting up the environment for using tinygrad.

```python
import numpy as np
from tinygrad.helpers import Timing
```

--------------------------------

### Install Data and ML Libraries (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.1/tinycorp/benchmarks/bert/implementations/tinybox_8xMI300X/README.md

Installs essential Python libraries for data handling and machine learning tasks. Includes gdown for dataset downloads, numpy for numerical operations, tqdm for progress bars, and tensorflow for machine learning.

```shell
pip install gdown numpy tqdm tensorflow
```

--------------------------------

### Install Data and ML Libraries (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_8xMI300X/README.md

Installs essential Python libraries for data handling and machine learning tasks. Includes gdown for dataset downloads, numpy for numerical operations, tqdm for progress bars, and tensorflow for machine learning.

```shell
pip install gdown numpy tqdm tensorflow
```

--------------------------------

### Install Data and ML Libraries (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.1/tinycorp/benchmarks/bert/implementations/tinybox_red/README.md

Installs essential Python libraries for data handling and machine learning tasks. Includes gdown for dataset downloads, numpy for numerical operations, tqdm for progress bars, and tensorflow for machine learning.

```shell
pip install gdown numpy tqdm tensorflow
```

--------------------------------

### Setup Snapdragon Profiler Scripts

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tools/profiling/snapdragon/README.md

Commands to set up the Snapdragon Profiler environment. This includes running the initial installation scripts and establishing a connection to the target device using ADB.

```shell
./setup-profiler.sh
./setup-agnos.sh
```

```shell
adb connect xxx
```

--------------------------------

### Install tinygrad and mlperf-logging (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.1/tinycorp/benchmarks/bert/implementations/tinybox_8xMI300X/README.md

Installs the tinygrad library and mlperf-logging from a specific branch. Requires cloning the tinygrad repository first. This is a prerequisite for using MLPerf features.

```shell
git clone https://github.com/tinygrad/tinygrad.git
python3 -m pip install -e ".[mlperf]"
```

--------------------------------

### Install Data and ML Libraries (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_green/README.md

Installs essential Python libraries for data handling and machine learning tasks. Includes gdown for dataset downloads, numpy for numerical operations, tqdm for progress bars, and tensorflow for machine learning.

```shell
pip install gdown numpy tqdm tensorflow
```

--------------------------------

### Install tinygrad and mlperf-logging (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_8xMI300X/README.md

Installs the tinygrad library and mlperf-logging from a specific branch. Requires cloning the tinygrad repository first. This is a prerequisite for using MLPerf features.

```shell
git clone https://github.com/tinygrad/tinygrad.git
python3 -m pip install -e ".[mlperf]"
```

--------------------------------

### Install Data and ML Libraries (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_red/README.md

Installs essential Python libraries for data handling and machine learning tasks. Includes gdown for dataset downloads, numpy for numerical operations, tqdm for progress bars, and tensorflow for machine learning.

```shell
pip install gdown numpy tqdm tensorflow
```

--------------------------------

### Install tinygrad and mlperf-logging (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.1/tinycorp/benchmarks/bert/implementations/tinybox_red/README.md

Installs the tinygrad library and mlperf-logging from a specific branch. Requires cloning the tinygrad repository first. This is a prerequisite for using MLPerf features.

```shell
git clone https://github.com/tinygrad/tinygrad.git
python3 -m pip install -e ".[mlperf]"
```

--------------------------------

### Install Data and ML Libraries (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.1/tinycorp/benchmarks/bert/implementations/tinybox_green/README.md

Installs essential Python libraries for data handling and machine learning tasks. Includes gdown for dataset downloads, numpy for numerical operations, tqdm for progress bars, and tensorflow for machine learning.

```shell
pip install gdown numpy tqdm tensorflow
```

--------------------------------

### Install tinygrad and mlperf-logging (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.1/tinycorp/benchmarks/bert/implementations/tinybox_green/README.md

Installs the tinygrad library and mlperf-logging from a specific branch. Requires cloning the tinygrad repository first. This is a prerequisite for using MLPerf features.

```shell
git clone https://github.com/tinygrad/tinygrad.git
python3 -m pip install -e ".[mlperf]"
```

--------------------------------

### Install tinygrad and mlperf-logging (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_red/README.md

Installs the tinygrad library and mlperf-logging from a specific branch. Requires cloning the tinygrad repository first. This is a prerequisite for using MLPerf features.

```shell
git clone https://github.com/tinygrad/tinygrad.git
python3 -m pip install -e ".[mlperf]"
```

--------------------------------

### Install tinygrad and mlperf-logging (Python)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/bert/implementations/tinybox_green/README.md

Installs the tinygrad library and mlperf-logging from a specific branch. Requires cloning the tinygrad repository first. This is a prerequisite for using MLPerf features.

```shell
git clone https://github.com/tinygrad/tinygrad.git
python3 -m pip install -e ".[mlperf]"
```

--------------------------------

### Launch Snapdragon Profiler

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tools/profiling/snapdragon/README.md

Steps to navigate to the Snapdragon Profiler directory and execute the main script to start the application. This allows you to connect to the device and begin profiling sessions.

```shell
cd SnapdragonProfiler
./run_sdp.sh
```

--------------------------------

### Install and Launch Jupyter Notebooks

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tools/car_porting/README.md

Instructions for installing Jupyter and its kernel within the openpilot virtual environment, and how to launch the Jupyter notebook server. This enables the use of provided example notebooks for data analysis.

```bash
uv pip install jupyter ipykernel

jupyter notebook
```

--------------------------------

### Project Setup and Testing

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/opendbc_repo/README.md

Commands to clone the repository, install dependencies, build the project, run tests, and lint the code. This sequence ensures the development environment is correctly set up and the project integrity is maintained.

```bash
git clone https://github.com/commaai/opendbc.git
cd opendbc
./test.sh
```

```bash
# Install dependencies, including testing and documentation tools
pip3 install -e .[testing,docs]
```

```bash
# Build the project using scons with parallel execution
scons -j8
```

```bash
# Run project tests using pytest
pytest .
```

```bash
# Run pre-commit hooks to lint and format all files
pre-commit run --all-files
```

--------------------------------

### Using openpilot in a Car Setup

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/README.md

Details the requirements for using openpilot in a vehicle, including the necessary hardware (comma 3/3X), software installation URL, car compatibility, and car harness.

```markdown
Using openpilot in a car
------

To use openpilot in a car, you need four things:
1. **Supported Device:** a comma 3/3X, available at [comma.ai/shop](https://comma.ai/shop/comma-3x).
2. **Software:** The setup procedure for the comma 3/3X allows users to enter a URL for custom software. Use the URL `openpilot.comma.ai` to install the release version.
3. **Supported Car:** Ensure that you have one of [the 275+ supported cars](docs/CARS.md).
4. **Car Harness:** You will also need a [car harness](https://comma.ai/shop/car-harness) to connect your comma 3/3X to your car.

We have detailed instructions for [how to install the harness and device in a car](https://comma.ai/setup). Note that it's possible to run openpilot on [other hardware](https://blog.comma.ai/self-driving-car-for-free/), although it's not plug-and-play.
```

--------------------------------

### One-time setup for tinybox_red

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_red/README.md

Performs the one-time setup required for the tinybox_red environment. This script configures the system before running benchmarks.

```shell
examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_red/setup.sh
```

--------------------------------

### Saving and Loading Model Weights with Safetensors

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/docs/quickstart.md

Provides an API reference for saving and loading neural network model weights using the safetensors format. It details the functions available in tinygrad.nn.state for managing model state dictionaries, including getting, saving, and loading.

```APIDOC
tinygrad.nn.state:
  - safe_save(state_dict: dict, filename: str):
    Saves a model's state dictionary to a file in safetensors format.
    Parameters:
      state_dict: A dictionary containing the model's weights and parameters.
      filename: The path to the file where the state dictionary will be saved.

  - safe_load(filename: str) -> dict:
    Loads a state dictionary from a safetensors file.
    Parameters:
      filename: The path to the safetensors file.
    Returns:
      A dictionary representing the loaded state dictionary.

  - get_state_dict(model: object) -> dict:
    Retrieves the state dictionary from a given model object.
    Parameters:
      model: The model object from which to extract the state dictionary.
    Returns:
      A dictionary containing the model's state.

  - load_state_dict(model: object, state_dict: dict):
    Loads a state dictionary into a model object.
    Parameters:
      model: The model object to load the state into.
      state_dict: The state dictionary to load.

Example Usage:
```python
from tinygrad.nn.state import safe_save, safe_load, get_state_dict, load_state_dict

# Assume 'net' is a pre-defined neural network model

# Get the state dictionary of the model
state_dict = get_state_dict(net)

# Save the state dictionary to a file
safe_save(state_dict, "model.safetensors")

# Load the state dictionary back from the file
loaded_state_dict = safe_load("model.safetensors")

# Load the state dictionary into the model
load_state_dict(net, loaded_state_dict)
```

Note: Many models in the tinygrad/extra/models folder provide a `load_from_pretrained` method, which often requires PyTorch to be installed for loading PyTorch weights.
```

--------------------------------

### PlotJuggler Installation and Setup

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tools/plotjuggler/README.md

Installs PlotJuggler and its openpilot plugins. Requires the openpilot environment to be set up beforehand.

```shell
cd tools/plotjuggler && ./juggle.py --install
```

--------------------------------

### One-time setup for tinybox_red

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_green/README.md

Performs the one-time setup required for the tinybox_red environment. This script configures the system before running benchmarks.

```shell
examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_red/setup.sh
```

--------------------------------

### Simulation Example Executable Target

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/third_party/acados/acados_template/c_templates_tera/CMakeLists.in.txt

Defines and installs a simulation example executable if BUILD_SIM_EXAMPLE is enabled. It links model and simulation object files.

```CMake
{% if solver_options.integrator_type != "DISCRETE" -%}
# example_sim
if(${BUILD_SIM_EXAMPLE})
    set(EX_SIM_SRC main_sim_{{ model.name }}.c)
    set(EX_SIM_EXE main_sim_{{ model.name }})
    add_executable(${EX_SIM_EXE} ${EX_SIM_SRC} $<TARGET_OBJECTS:${MODEL_OBJ}> $<TARGET_OBJECTS:${SIM_OBJ}>)
    install(TARGETS ${EX_SIM_EXE} DESTINATION ${CMAKE_INSTALL_PREFIX})
endif(${BUILD_SIM_EXAMPLE})
{%- endif %}
```

--------------------------------

### Example: Full Camera Streaming Workflow

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tools/camerastream/README.md

An example demonstrating the complete workflow: starting the stream decoder on the PC and then launching the display utility. This shows the typical usage pattern.

```bash
cd ~/openpilot/tools/camerastream && ./compressed_vipc.py comma-ffffffff --cams 0
cd ~/openpilot/selfdrive/ui/ && ./watch3
```

--------------------------------

### Deploy Fork to Device

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/docs/how-to/turn-the-speed-blue.md

Provides the URL format to install the custom openpilot fork onto a comma device. This involves uninstalling the existing version and then accessing a personalized installer URL.

```bash
installer.comma.ai/<your-github-username>/master
```

--------------------------------

### Serve openpilot docs locally (bash)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/docs/README.md

Starts a local web server to preview the openpilot documentation. The site will typically be accessible via http://localhost:8000. Run from the root directory of the openpilot project.

```bash
mkdocs serve
```

--------------------------------

### Stochastic Gradient Descent (SGD) Optimizer Initialization

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/docs/quickstart.md

Demonstrates how to initialize the Stochastic Gradient Descent (SGD) optimizer in tinygrad. It shows passing the network's parameters and setting a learning rate, with a note on using `get_parameters` for a more concise approach.

```python
from tinygrad.nn.optim import SGD

opt = SGD([net.l1.weight, net.l2.weight], lr=3e-4)
```

--------------------------------

### Install Additional Dependencies

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.1/tinycorp/benchmarks/retinanet/implementations/tinybox_green/README.md

Installs essential Python packages required for the project, including data handling, machine learning, and utility libraries.

```shell
pip install tqdm numpy pycocotools boto3 pandas torch torchvision
```

--------------------------------

### Device Setup Commands

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tools/camerastream/README.md

Commands to run on the openpilot device to start the necessary daemons for streaming video. These include bridge, encorderd, and camerad.

```bash
cd /data/openpilot/cereal/messaging && ./bridge
cd /data/openpilot/system/loggerd && ./encoderd
cd /data/openpilot/system/camerad && ./camerad
```

```bash
(
  cd /data/openpilot/cereal/messaging/
  ./bridge &

  cd /data/openpilot/system/camerad/
  ./camerad &

  cd /data/openpilot/system/loggerd/
  ./encoderd &

  wait
) ; trap 'kill $(jobs -p)' SIGINT
```

--------------------------------

### Tensor Creation with Data Types

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/docs/quickstart.md

Shows how to specify the data type (dtype) for Tensors during creation, using predefined dtypes from the tinygrad library.

```python
from tinygrad import Tensor, dtypes

t3 = Tensor([1, 2, 3, 4, 5], dtype=dtypes.int32)
```

--------------------------------

### Install Additional Dependencies

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/retinanet/implementations/tinybox_green/README.md

Installs essential Python packages required for the project, including data handling, machine learning, and utility libraries.

```shell
pip install tqdm numpy pycocotools boto3 pandas torch torchvision
```

--------------------------------

### Tensor Creation from Data Structures

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/docs/quickstart.md

Demonstrates how to create tinygrad Tensors from Python lists and NumPy ndarrays, serving as the fundamental data structure for operations.

```python
from tinygrad import Tensor

t1 = Tensor([1, 2, 3, 4, 5])
na = np.array([1, 2, 3, 4, 5])
t2 = Tensor(na)
```

--------------------------------

### Install tinygrad from Source

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/docs/index.md

Instructions for cloning the tinygrad repository and installing it locally using pip. This method is recommended for users who want the latest features or to contribute to the project.

```bash
git clone https://github.com/tinygrad/tinygrad.git
cd tinygrad
python3 -m pip install -e .
```

--------------------------------

### Install Tinygrad and MLPerf

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.0/tinycorp/benchmarks/retinanet/implementations/tinybox_green/README.md

Installs the tinygrad library and mlperf-logging from a specific branch. This involves cloning the repository and performing a local Python package installation.

```shell
git clone https://github.com/tinygrad/tinygrad.git
python3 -m pip install -e ".[mlperf]"
```

--------------------------------

### Basic Tensor Operations and Realization

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/docs/quickstart.md

Demonstrates performing arithmetic operations on Tensors and applying activation functions. Operations are lazy and executed upon calling `.realize()` or `.numpy()`.

```python
from tinygrad import Tensor

t4 = Tensor([1, 2, 3, 4, 5])
t5 = (t4 + 1) * 2
t6 = (t5 * t4).relu().log_softmax()

print(t6.numpy())
# [-56. -48. -36. -20.   0.]
```

--------------------------------

### Download Wikipedia Raw Data (Shell)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/examples/mlperf/training_submission_v5.1/tinycorp/benchmarks/bert/implementations/tinybox_red/README.md

Downloads the raw Wikipedia dataset. Sets the base directory for datasets and enables verification of checksums. This script is part of the data preparation steps.

```shell
BASEDIR="/raid/datasets/wiki" WIKI_TRAIN=1 VERIFY_CHECKSUM=1 python3 extra/datasets/wikipedia_download.py
```

--------------------------------

### Clone openpilot Repository (Full)

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tools/README.md

Clones the complete openpilot repository, including all history and data, using Git LFS. This method requires Git LFS to be installed and configured.

```bash
git clone --recurse-submodules https://github.com/commaai/openpilot.git
```

--------------------------------

### MNIST Dataset Fetching

Source: https://github.com/ajouatom/openpilot/blob/carrot2-v8/tinygrad_repo/docs/quickstart.md

Imports the MNIST dataset loader from the `extra.datasets` module within tinygrad. This function is used to retrieve the MNIST training and testing data for model evaluation.

```python
from extra.datasets import fetch_mnist
```