### Install 3LC and 3LC Examples

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/2-modify-tables/1-weight-coreset.html

Installs the necessary 3LC libraries and example code from GitHub. This is a prerequisite for running the subsequent code examples.

```bash
%pip install 3lc
%pip install git+https://github.com/3lc-ai/3lc-examples.git
```

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

### Install Dependencies for 3LC Examples

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/create-semantic-segmentation-table.html

Installs necessary libraries for running 3LC examples, including the 3LC library itself, example-specific repositories, and Hugging Face utilities.

```python
%pip install 3lc
%pip install git+https://github.com/3lc-ai/3lc-examples.git
%pip install huggingface-hub
```

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

### Install 3LC and Hugging Face dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/huggingface-coco128-segformer-collection.html

Installs the necessary 3LC packages and example tools required to run the segmentation workflow.

```bash
%pip install 3lc[huggingface]
%pip install git+https://github.com/3lc-ai/3lc-examples.git
```

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

### Install 3LC Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/2-modify-tables/add-instance-embeddings/1-train-crop-model.html

Installs the necessary 3LC libraries, example tools, and the timm library for model architectures.

```bash
%pip install 3lc[pacmap]
%pip install git+https://github.com/3lc-ai/3lc-examples.git
%pip install timm
```

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

### Install 3LC and dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/pytorch-cifar10-collect-metrics.html

Installs the required 3LC library, timm for model architectures, and example utilities via pip.

```bash
%pip install 3lc
%pip install timm
%pip install git+https://github.com/3lc-ai/3lc-examples.git
```

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

### Install 3LC and Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/2-modify-tables/split-tables.html

Installs the 3LC library, the 3LC examples repository, and Seaborn for data visualization. These are essential dependencies for running the provided examples.

```python
%pip install 3lc
%pip install git+https://github.com/3lc-ai/3lc-examples.git
%pip install seaborn

```

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

### Start 3LC Object Service

Source: https://docs.3lc.ai/3lc/latest/getting-started/index.html

Launch the 3LC Object Service. Options include running with a text-based UI, disabling it, or disabling public examples.

```bash
3lc service
```

```bash
3lc service --no-tui
```

```bash
3lc service --no-public-examples
```

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

### Install 3LC and Hugging Face dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/huggingface-ade20k-segformer-finetuning.html

Installs the required 3LC library with Hugging Face support and the necessary example utilities via pip.

```bash
%pip install 3lc[huggingface] "transformers<=4.56.0"
%pip install git+https://github.com/3lc-ai/3lc-examples.git
```

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

### Initialize 3LC Project and Install Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/keypoints/create-custom-keypoints-table.html

Sets up the environment variables for the 3LC project and installs necessary packages including the 3LC SDK and kagglehub for dataset retrieval.

```python
PROJECT_NAME = "3LC Tutorials - 2D Keypoints"
DATASET_NAME = "AnimalPose"
TABLE_NAME = "initial"
DOWNLOAD_PATH = "../../transient_data"

%pip install 3lc
%pip install kagglehub
```

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

### Install 3LC and SAM dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/2-modify-tables/transform-bbs-to-segs.html

Installs the necessary Python packages including 3LC, the 3LC examples repository, the Segment Anything Model, and matplotlib for visualization.

```bash
%pip install 3lc
%pip install git+https://github.com/3lc-ai/3lc-examples.git
%pip install git+https://github.com/facebookresearch/segment-anything
%pip install matplotlib
```

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

### Install 3LC Library

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/create-fhibe-table.html

Installs the 3LC library using pip. This is a prerequisite for running the subsequent code in the notebook.

```bash
%pip install -q 3lc

```

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

### Setup and Install 3LC Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/instance-segmentation/create-table-from-coco-segmentation.html

Configures the environment variables for the 3LC project and installs the necessary Python package.

```python
PROJECT_NAME = "3LC Tutorials - COCO128"
DATASET_NAME = "COCO128"
TABLE_NAME = "initial-segmentation"
DATA_PATH = "../../../data"

%pip install 3lc
```

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

### Install 3LC Library

Source: https://docs.3lc.ai/3lc/latest/getting-started/deployment-examples/remote-training-local-service-s3.html

Installs the 3LC Python library using pip. This is the first step for both training VMs and team users.

```bash
pip install 3lc
```

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

### Configure Project Environment and Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/instance-segmentation/create-custom-instance-segmentation-table.html

Initializes project constants and installs necessary libraries including 3LC, matplotlib, and kaggle API tools. This setup is required to manage data paths and enable 3LC functionality.

```python
PROJECT_NAME = "3LC Tutorials - Cell Segmentation"
DATASET_NAME = "Sartorius Cell Segmentation"
TABLE_NAME = "initial"
DOWNLOAD_PATH = "../../../transient_data"

%pip install -q 3lc
%pip install -q matplotlib
%pip install -q kaggle
%pip install -q git+https://github.com/3lc-ai/3lc-examples.git
```

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

### Install Required Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/lightning-balloons-segformer-training.html

Installs the necessary 3LC, PyTorch Lightning, and visualization libraries via pip.

```bash
%pip install 3lc[huggingface]
%pip install pytorch-lightning
%pip install matplotlib
```

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

### Configure 3LC Project and Install Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/pytorch-augmentation-explorer.html

Sets up the environment variables for the 3LC project and installs the necessary package via pip.

```python
PROJECT_NAME = "3LC Tutorials - Augmentation Explorer"
DATASET_NAME = "COCO128"
TABLE_NAME = "images-only"
RUN_NAME = "register-augmented-samples"
RUN_DESCRIPTION = "Inspecting augmentations on COCO-128"
DATA_PATH = "../../data"
BATCH_SIZE = 32
EPOCHS = 10
INSTALL_DEPENDENCIES = True

if INSTALL_DEPENDENCIES:
    %pip install 3lc
```

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

### Initialize Project and Install Dependencies for 3LC

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/obbs/create-table-from-yolo-obb.html

Sets up the project configuration variables and installs the required 3LC library. This is the prerequisite step for interacting with 3LC tables.

```python
PROJECT_NAME = "3LC Tutorials - YOLO OBB"
DATASET_NAME = "YOLO-OBB-Dataset"
TABLE_NAME = "initial-obb"
DATA_PATH = "../../../data"

%pip install 3lc

import tlc
```

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

### Setup and Train YOLO Pose Estimation Model

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/ultralytics-animalpose-yolo11-pose-training.html

Configures the project environment, installs necessary 3LC and Ultralytics dependencies, performs stratified data splitting on keypoint tables, and executes the training process.

```python
PROJECT_NAME = "3LC Tutorials - 2D Keypoints"
RUN_NAME = "train-yolon-animalpose"
RUN_DESCRIPTION = "Train yolo11n-pose on the AnimalPose dataset"
DATASET_NAME = "AnimalPose"
TABLE_NAME = "initial"
NUM_WORKERS = 0
EPOCHS = 10

%pip install 3lc-ultralytics
%pip install git+https://github.com/3lc-ai/3lc-examples.git

import tlc
from tlc_ultralytics import YOLO, Settings
from tlc_tools.split import split_table

initial_table = tlc.Table.from_names(TABLE_NAME, DATASET_NAME, PROJECT_NAME)

def split_by(table_row):
    return table_row["keypoints_2d"]["instances_additional_data"]["label"][0]

train_val_test = split_table(
    initial_table,
    splits={"train": 0.8, "val_test": 0.2},
    split_strategy="stratified",
    split_by=split_by,
    random_seed=42,
    shuffle=False,
)

model = YOLO("yolo11n-pose.pt")
settings = Settings(
    project_name=PROJECT_NAME,
    run_name=RUN_NAME,
    run_description=RUN_DESCRIPTION,
    collect_loss=True,
    image_embeddings_dim=2,
)

model.train(
    tables={"train": train_val_test["train"], "val": train_val_test["val_test"]},
    epochs=EPOCHS,
    workers=NUM_WORKERS,
)
```

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

### Install Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/obbs/create-custom-obb-table.html

Installs the necessary Python packages, `gdown` for downloading the dataset and `3lc` for data manipulation and model training.

```python
%pip install -q gdown
%pip install -q 3lc

```

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

### Install 3LC with NGrok Support

Source: https://docs.3lc.ai/3lc/latest/python-package/object-service/object-service-deployment-guide.html

Installs the 3LC Python package including the necessary dependencies for NGrok integration.

```bash
pip install 3lc[pyngrok]
```

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

### Install 3LC Package

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/3d/write-mammoth-table.html

Installs the 3LC package using pip. This is a prerequisite for using 3LC functionalities in the notebook.

```python
if INSTALL_DEPENDENCIES:
    %pip --quiet install 3lc
```

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

### Configure 3LC Project and Environment

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/huggingface-cifar100-collect-embeddings.html

Sets up the project metadata, model parameters, and installs necessary dependencies including 3LC, UMAP, and Hugging Face transformers.

```python
PROJECT_NAME = "3LC Tutorials - CIFAR-100"
RUN_NAME = "Collect Image Embeddings"
DESCRIPTION = "Collect image embeddings from ViT model on CIFAR-100"
DEVICE = None
TRAIN_DATASET_NAME = "hf-cifar-100-train"
TEST_DATASET_NAME = "hf-cifar-100-test"
MODEL = "google/vit-base-patch16-224"
BATCH_SIZE = 32
NUM_WORKERS = 4
INSTALL_DEPENDENCIES = True

if INSTALL_DEPENDENCIES:
    %pip -q install 3lc[umap,huggingface] "transformers<=4.56.0"
```

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

### Install SuperGradients and 3LC

Source: https://docs.3lc.ai/3lc/latest/python-package/integrations/supergradients/supergradients.html

Installs the SuperGradients library and the 3LC library. It addresses potential version conflicts with the 'termcolor' package by recommending separate installation steps.

```bash
pip install super-gradients
pip install 3lc
```

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

### Install Dependencies for 3LC and Detectron2

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/detectron2-balloons-detection-finetuning.html

Installs the necessary libraries including PyTorch, Detectron2, and the 3LC SDK. Note that specific versions are required based on the system architecture and CUDA version.

```python
if INSTALL_DEPENDENCIES:
    %pip install torch==1.10.1+cu111 torchvision==0.11.2+cu111 -f https://download.pytorch.org/whl/cu111/torch_stable.html
    %pip install detectron2 -f "https://dl.fbaipublicfiles.com/detectron2/wheels/cu111/torch1.10/index.html"
    %pip install 3lc
    %pip install opencv-python
    %pip install matplotlib
    %pip install numpy==1.24.4
```

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

### Install Dependencies for Detectron2 and 3LC

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/detectron2-coco128-detection-collection.html

Installs the required versions of PyTorch, Detectron2, 3LC, and utility libraries for the environment.

```bash
%pip install --force-reinstall torch==1.10.1+cu111 torchvision==0.11.2+cu111 -f https://download.pytorch.org/whl/cu111/torch_stable.html
%pip install detectron2 -f "https://dl.fbaipublicfiles.com/detectron2/wheels/cu111/torch1.10/index.html"
%pip install 3lc[pacmap]
%pip install opencv-python
%pip install matplotlib
%pip install numpy==1.24.4
```

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

### Configure Project and Install Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/keypoints/create-table-from-coco-keypoints.html

Sets up the project environment variables and installs the required 3LC Python package to enable data table creation.

```python
PROJECT_NAME = "3LC Tutorials - COCO128"
DATASET_NAME = "COCO128"
TABLE_NAME = "initial-keypoints"
DATA_PATH = "../../../data"

%pip install 3lc
```

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

### Install Required Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/detectron2-coco128-segmentation-collection.html

Installs the necessary libraries including PyTorch, Detectron2, 3LC, and utility packages like OpenCV and Matplotlib to support the metrics collection pipeline.

```python
if INSTALL_DEPENDENCIES:
    %pip install torch==1.10.1+cu111 torchvision==0.11.2+cu111 -f https://download.pytorch.org/whl/cu111/torch_stable.html
    %pip install detectron2 -f "https://dl.fbaipublicfiles.com/detectron2/wheels/cu111/torch1.10/index.html"
    %pip install 3lc[pacmap]
    %pip install opencv-python
    %pip install matplotlib
    %pip install numpy==1.24.4
```

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

### Install dependencies for 3LC and SuperGradients

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/supergradients-animalpose-yolonas-training.html

Installs the necessary Python packages including 3LC, super-gradients, and specific versions of dependencies required for the pose estimation workflow.

```bash
%pip install 3lc
%pip install super-gradients
%pip install termcolor==3.1.0
%pip install git+https://github.com/3lc-ai/3lc-examples.git
```

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

### Install dependencies for 3LC segmentation workflows

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/instance-segmentation/create-instance-segmentations-from-image-masks.html

Installs the necessary Python packages including the 3LC SDK, Hugging Face Hub, and utility libraries required for data processing and visualization.

```bash
%pip install 3lc
%pip install huggingface-hub
%pip install git+https://github.com/3lc-ai/3lc-examples.git
%pip install matplotlib
```

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

### Install Detectron2 and 3LC Dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/detectron2-balloons-segmentation-finetuning.html

Installs necessary libraries including PyTorch with CUDA support, Detectron2, the 3LC library, OpenCV, Matplotlib, and NumPy. This step ensures all required tools are available for the training process.

```python
%pip install torch==1.10.1+cu111 torchvision==0.11.2+cu111 -f https://download.pytorch.org/whl/cu111/torch_stable.html
%pip install detectron2 -f "https://dl.fbaipublicfiles.com/detectron2/wheels/cu111/torch1.10/index.html"
%pip install 3lc
%pip install opencv-python
%pip install matplotlib
%pip install numpy==1.24.4

```

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

### Initialize and Use the Url Class

Source: https://docs.3lc.ai/3lc/latest/python-package/apidocs/tlcurl/tlcurl.url.html

Demonstrates how to instantiate the Url class with various schemes and paths. It shows how the class handles automatic scheme detection and alias expansion.

```python
file_url = Url("/path/to/file")
assert file_url.scheme == Scheme.FILE

s3_url = Url("s3://bucket/path/to/object")
assert s3_url.scheme == Scheme.S3

# Aliases are expanded when the URL is used
alias_url = Url("<SAMPLE_DATA>/data.csv")
UrlAliasRegistry.instance().register_url_alias(token="<SAMPLE_DATA>", path="/path/to/data")
```

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

### Initialize and Access OptionLoader Instance

Source: https://docs.3lc.ai/3lc/latest/python-package/apidocs/tlcconfig/tlcconfig.option_loader.html

Demonstrates how to instantiate the OptionLoader and retrieve the global singleton instance for the application.

```python
from tlc.option_loader import OptionLoader

# Initialize a new loader
loader = OptionLoader(config_file="config.yaml", initialize=True)

# Access the global singleton instance
instance = OptionLoader.instance()
```

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

### Create 3LC Image Tables using various schemas

Source: https://docs.3lc.ai/3lc/latest/user-guide/column-types/computer-vision/image.html

Demonstrates how to initialize a 3LC table with image data. Includes examples using built-in URL schemas, explicit schema definitions, and direct PIL Image object integration.

```python
import tlc

# Using Builtin schema
table = tlc.Table.from_dict(
    data={"image": ["path/to/image0.png", "path/to/image1.png", ...]},
    structure={"image": tlc.ImageUrlSchema()},
)
```

```python
import tlc

# Using explicit Schema definition
table = tlc.Table.from_dict(
    data={"image": ["path/to/image0.png", "path/to/image1.png", ...]},
    structure={"image": tlc.Schema(value=tlc.StringValue(string_role=tlc.STRING_ROLE_IMAGE_URL))},
)
```

```python
import tlc
from PIL import Image

# Using PIL Image objects
table = tlc.Table.from_dict(
    data={"image": [Image.open("path/to/image0.png"), Image.open("path/to/image1.png"), ...]},
    structure={"image": tlc.PILImage},
)
```

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

### Run Training with Main.py Script

Source: https://docs.3lc.ai/3lc/latest/python-package/integrations/supergradients/supergradients.html

This command demonstrates how to initiate the training process using the `main.py` script with a specified YAML recipe configuration file.

```bash
python main.py --config-name=my_recipe.yaml


```

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

### Install 3LC and dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/2-modify-tables/add-embeddings.html

Installs the necessary Python packages including 3LC with PaCMAP support, example tools, and the Hugging Face transformers library.

```bash
%pip install 3lc[pacmap]
%pip install git+https://github.com/3lc-ai/3lc-examples.git
%pip install transformers
```

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

### Get URL Extension

Source: https://docs.3lc.ai/3lc/latest/python-package/apidocs/tlcurl/tlcurl.url.html

Retrieves the file extension from the URL. For example, 'example.json' would return '.json'.

```python
property extension: str
    """
    Get the extension of the URL.
    Example:
    ```
    Url("example.json").extension == ".json"
    ```

    Returns:
        The extension of the URL.
    """
    pass
```

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

### Install 3LC with UMAP and PACMAP dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/2-modify-tables/dimensionality-reduction-toy-example.html

Installs the 3LC library with necessary dependencies for UMAP and PACMAP dimensionality reduction algorithms. This command ensures that all required packages are available for the subsequent code examples.

```bash
%pip install "3lc[umap,pacmap]"
```

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

### Configuration: System Options

Source: https://docs.3lc.ai/3lc/latest/python-package/apidocs/tlcconfig/tlcconfig.options.html

General system configuration options like display progress and logging.

```APIDOC
## Configuration: System Options

### Description
Settings for controlling system behavior such as progress bar visibility and log file paths.

### Keys
- `tlc.display-progress` (int) - 0 or 1, toggles progress bars.
- `logfile` (string) - Path to the system log file.

### Request Example
{
  "tlc.display-progress": 1,
  "logfile": "/var/log/3lc/system.log"
}
```

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

### Get URL Name

Source: https://docs.3lc.ai/3lc/latest/python-package/apidocs/tlcurl/tlcurl.url.html

Retrieves the name component of the URL. This could be a filename or a directory name. For example, 'C:/folder/file.txt' returns 'file.txt', and 'C:/folder' returns 'folder'.

```python
property name: str
    """
    Get the name of the URL.
    Example:
    ```
    Url("C:/folder/file.txt").name == "file.txt"
    Url("C:/folder").name == "folder"
    ```

    Returns:
        The name of the URL.
    """
    pass
```

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

### Register MetricsCollectionHook in Detectron2 Trainer

Source: https://docs.3lc.ai/3lc/latest/python-package/apidocs/tlc/tlc.integration.detectron2.metrics_collection_hook.html

Demonstrates how to instantiate and register the MetricsCollectionHook within a Detectron2 training workflow. This setup enables automated metric collection on a specified dataset starting at a defined iteration.

```python
test_hook = MetricsCollectionHook(
    dataset_name="my_test_dataset",
    metrics_collectors=MyMetricsCollector(),
    collection_start_iteration=1000,
    collection_frequency=100,
)

trainer.register_hooks([train_hook, test_hook])
trainer.train()
```

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

### Manage Configuration Values

Source: https://docs.3lc.ai/3lc/latest/python-package/apidocs/tlcconfig/tlcconfig.option_loader.html

Shows how to retrieve and update configuration options programmatically using the OptionLoader instance.

```python
loader = OptionLoader.instance()

# Get a specific configuration value
value = loader.get_value(MyOptionClass)

# Set a configuration value
loader.set_value(MyOptionClass, "new_value")

# Validate the current configuration state
loader.validate()
```

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

### Export Configuration to YAML

Source: https://docs.3lc.ai/3lc/latest/python-package/apidocs/tlcconfig/tlcconfig.option_loader.html

Provides examples for serializing the current configuration state into YAML format for storage or inspection.

```python
loader = OptionLoader.instance()

# Get YAML string representation
yaml_str = loader.to_yaml(doc_strings=True, detail=True)

# Write current configuration to a file
loader.write_to_yaml_file("settings.yaml", overwrite=True)
```

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

### Install 3LC Library (Shell)

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/create-table-from-image-folder.html

This command installs the 3LC library using pip. Ensure you have pip installed and configured correctly in your environment before running this command.

```shell
%pip install 3lc
```

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

### Launch 3LC Object Service with NGrok

Source: https://docs.3lc.ai/3lc/latest/python-package/object-service/object-service-deployment-guide.html

Starts the 3LC Object Service with the --ngrok flag to automatically generate a public URL for remote access.

```bash
3lc service --ngrok
```

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

### Example ID Schema

Source: https://docs.3lc.ai/3lc/latest/python-package/apidocs/tlc/tlc.core.builtins.schemas.schemas.html

Schema for representing unique example identifiers.

```APIDOC
## ExampleIdSchema

### Description
A schema for example ID values. Example ID is a unique identifier for an example. It is used to identify examples across different tables.

### Method
N/A (Schema Definition)

### Endpoint
N/A

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
* **display_name** (str) - Default: 'Example ID'
* **description** (str) - Default: ''
* **writable** (bool) - Default: False
* **computable** (bool) - Default: False

### Request Example
```json
{
  "display_name": "Example ID",
  "description": "",
  "writable": false,
  "computable": false
}
```

### Response
#### Success Response (200)
* **Schema Definition** - Represents an example ID.

#### Response Example
```json
{
  "example": "ExampleIdSchema"
}
```
```

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

### Initialize 3LC Training Environment

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/2-modify-tables/add-instance-embeddings/1-train-crop-model.html

Imports required modules, sets configuration constants, and detects the available hardware device for training.

```python
import tlc
from tlc_tools.augment_bbs.finetune_on_crops import train_model
from tlc_tools.common import infer_torch_device
from tlc_tools.split import split_table

EPOCHS = 10
TMP_PATH = "../../../transient_data"
MODEL_NAME = "efficientnet_b0"
NUM_WORKERS = 0
MODEL_CHECKPOINT = TMP_PATH + "/instance_classifier.pth"

DEVICE = infer_torch_device()
print(f"Using device: {DEVICE}")
```

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

### Setup Detectron2 Project Configuration

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/detectron2-balloons-segmentation-finetuning.html

Defines project-specific configurations for fine-tuning a Detectron2 instance segmentation model, including dataset names, paths, model configuration, and training parameters. These variables are essential for setting up the training environment.

```python
PROJECT_NAME = "3LC Tutorials - Balloons"
RUN_NAME = "Fine-tune segmentation model"
DESCRIPTION = "Train a balloon segmenter using detectron2"
TRAIN_DATASET_NAME = "balloons-train-seg"
VAL_DATASET_NAME = "balloons-val-seg"
TMP_PATH = "../transient_data"
DATA_PATH = "../../data"
MODEL_CONFIG = "COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml"
MAX_ITERS = 200
BATCH_SIZE = 2
MAX_DETECTIONS_PER_IMAGE = 30
SCORE_THRESH_TEST = 0.5
MASK_FORMAT = "bitmask"

```

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

### Create Table and Run URLs

Source: https://docs.3lc.ai/3lc/latest/python-package/urls.html

Demonstrates how to generate standard URLs for tables and runs using the tlc.Url utility methods. These methods require parameters like table/run names, dataset names, and project names to construct the correct path.

```python
import tlc

table_url = tlc.Url.create_table_url(table_name, dataset_name, project_name)
run_url = tlc.Url.create_run_url(run_name, project_name)
```

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

### Install 3LC dependency

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/create-table-from-torch-dataset.html

Installs the 3LC library using pip to enable dataset conversion functionality.

```bash
%pip install -q 3lc
```

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

### Initialize 3LC run and configure Detectron2

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/detectron2-balloons-detection-finetuning.html

Initializes a 3LC tracking run and sets up the Detectron2 configuration object with model parameters and solver settings.

```python
run = tlc.init(
    project_name=PROJECT_NAME,
    run_name=RUN_NAME,
    description=DESCRIPTION,
    if_exists="overwrite",
)

cfg = get_cfg()
cfg.merge_from_file(model_zoo.get_config_file(MODEL_CONFIG))
cfg.DATASETS.TRAIN = (TRAIN_DATASET_NAME,)
cfg.DATASETS.TEST = (VAL_DATASET_NAME,)
cfg.MODEL.WEIGHTS = model_zoo.get_checkpoint_url(MODEL_CONFIG)
cfg.SOLVER.BASE_LR = 0.00025
cfg.MODEL.DEVICE = "cuda"
os.makedirs(cfg.OUTPUT_DIR, exist_ok=True)
```

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

### Initialize 3LC run and configure Detectron2

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/detectron2-coco128-detection-collection.html

Initializes a 3LC run and sets up the Detectron2 configuration object with model parameters and training settings.

```python
run = tlc.init(
    PROJECT_NAME,
    run_name=RUN_NAME,
    description=DESCRIPTION,
    if_exists="overwrite",
)

cfg = get_cfg()
cfg.merge_from_file(model_zoo.get_config_file(MODEL_CONFIG))
cfg.MODEL.WEIGHTS = model_zoo.get_checkpoint_url(MODEL_CONFIG)
cfg.DATASETS.TRAIN = (TRAIN_DATASET_NAME,)
cfg.OUTPUT_DIR = TMP_PATH
cfg.DATALOADER.NUM_WORKERS = 0
cfg.MODEL.ROI_HEADS.BATCH_SIZE_PER_IMAGE = 512
cfg.MODEL.ROI_HEADS.NUM_CLASSES = 80
cfg.TEST.DETECTIONS_PER_IMAGE = MAX_DETECTIONS_PER_IMAGE
cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = SCORE_THRESH_TEST
cfg.MODEL.DEVICE = "cuda"
cfg.DATALOADER.FILTER_EMPTY_ANNOTATIONS = False

config = {
    "model_config": MODEL_CONFIG,
    "test.detections_per_image": MAX_DETECTIONS_PER_IMAGE,
    "model.roi_heads.score_thresh_test": SCORE_THRESH_TEST,
}

run.set_parameters(config)
```

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

### Configure and Initialize 3LC Training Environment

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/3-training-and-metrics/huggingface-mrcp-bert-finetuning.html

Sets up global configuration variables, installs necessary dependencies, and initializes a 3LC run to track the fine-tuning process.

```python
PROJECT_NAME = "3LC Tutorials - Hugging Face BERT"
RUN_NAME = "finetuning-run"
DESCRIPTION = "Fine-tune BERT on MRPC"
TRAIN_DATASET_NAME = "hugging-face-train"
VAL_DATASET_NAME = "hugging-face-val"
CHECKPOINT = "bert-base-uncased"

# Install dependencies
%pip install accelerate scikit-learn 3lc[huggingface] "transformers<=4.56.0"

# Initialize 3LC Run
import tlc
run = tlc.init(
    project_name=PROJECT_NAME,
    run_name=RUN_NAME,
    description=DESCRIPTION,
    if_exists="overwrite"
)
```

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

### Convert starts and lengths to COCO RLE

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/instance-segmentation/create-custom-instance-segmentation-table.html

Converts a list of starts and lengths into a binary mask and encodes it into the COCO RLE format.

```python
def starts_lengths_to_coco_rle(starts_lengths, image_height, image_width):
    s = np.array(starts_lengths, dtype=int)
    starts = s[0::2] - 1
    lengths = s[1::2]

    mask = np.zeros(image_height * image_width, dtype=np.uint8)
    for start, length in zip(starts, lengths):
        mask[start : start + length] = 1
    mask = mask.reshape(image_height, image_width)

    rle = mask_utils.encode(np.asfortranarray(mask))
    return rle["counts"].decode("utf-8")
```

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

### Install 3LC and dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/2-modify-tables/add-instance-embeddings/2-add-instance-metrics.html

Installs the required 3LC packages, including PaCMAP for dimensionality reduction, along with utility libraries and timm for model support.

```bash
%pip install -q 3lc[pacmap]
%pip install -q git+https://github.com/3lc-ai/3lc-examples.git
%pip install -q timm
```

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

### Initialize 3LC Integration for Recipes

Source: https://docs.3lc.ai/3lc/latest/python-package/integrations/supergradients/supergradients.html

Demonstrates how to import the 3LC integration module in a main entrypoint script to register datasets and callbacks for use in SuperGradients YAML recipes.

```python
import omegaconf
import hydra
from super_gradients import Trainer, init_trainer
```

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

### Install 3LC and dependencies

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/2-modify-tables/add-classification-metrics.html

Installs the necessary Python packages including 3LC, transformers, and dimensionality reduction tools required for metric computation.

```bash
%pip install 3lc
%pip install git+https://github.com/3lc-ai/3lc-examples.git
%pip install transformers
%pip install pacmap
%pip install joblib
```

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

### Install 3LC Library

Source: https://docs.3lc.ai/3lc/latest/examples/tutorials/1-create-tables/create-custom-table.html

Installs the 3LC library, which is a dependency for creating and managing custom tables and datasets. This is typically the first step before utilizing any 3LC functionalities.

```bash
%pip install 3lc

```