Your submission must be a **CSV** with **two columns**:

1. **`image_id`** --> This must match the test-image metadata and must stay **exactly as provided** (same values and order) in `sample_submission.csv`.

2. **`cluster`**  --> Your predicted identity label for each image. Cluster names must follow one of these patterns:

   - `cluster_LynxID2025_???`
   - `cluster_SalamanderID2025_???`
   - `cluster_SeaTurtleID2022_???`
   - `cluster_TexasHornedLizards_???`

Here `???` can be any identifier you choose (e.g., `0`, `1`, `2`, …), as long as images you believe belong to the **same individual** share the **same** cluster name.

**Example (header required):**
```csv
image_id,cluster
3,cluster_LynxID2025_71
5,cluster_LynxID2025_0
12,cluster_LynxID2025_72
....
```