This plugin uses Ultralytics YOLO, which is dual-licensed: free under the Ultralytics AGPL-3.0 license for use that complies with its open-source terms, while commercial use beyond those terms requires a paid Ultralytics Enterprise License. It is your responsibility to ensure your use is appropriately licensed.

Ultralytics YOLO Training

Train Ultralytics YOLO models on your 3LC tables with automatic task detection, per-sample loss tracking, and embedding collection for data debugging and fixing.

Auto-detect

Task type detected from your table schema automatically

Per-sample loss

Track individual sample losses to find mislabeled or hard examples

Embeddings

Collect embeddings for visual clustering in the Dashboard

Collect mode

Run inference-only to collect metrics without training

1 Select tables

2 Configure model

3 Train

Config

1 Input Tables

Config Name

Train Table URL

Select your training data table. The task type (detection, segmentation, classification) is auto-detected from the table schema.

Validation Table URL (optional — validation skipped if empty)

Optional validation set for tracking metrics during training. If empty, validation is skipped (val=False).

Use .latest() at train time (resolve to newest table revision)

Detected Task

2 Ultralytics YOLO Settings

Model

Choose a YOLO model size. Larger models are more accurate but slower to train. Nano (n) is great for testing.

Mode

Train a YOLO model with 3LC metrics collection.

Pretrained Model URL (optional — fine-tune from existing checkpoint)

Copy best model to run folder (store checkpoint alongside run data)

Worker Threads

Number of dataloader worker threads. 0 = main thread only. Higher values speed up data loading on multi-core machines.

3 3LC Settings

3LC Project

Run Name

Queue & Progress

No active job.

Log