Rename a Workflow

Why rename workflows?

Clear, descriptive workflow names help you:

• Identify purpose — Quickly understand what each workflow trains
• Organize configurations — Distinguish between experimental and production workflows
• Track experiments — Use naming conventions for systematic testing
• Collaborate effectively — Help team members understand workflow objectives

Example naming conventions:

Rename a workflow

Step 1: Open your workflow

Navigate to the Training project containing your workflow.

From the Workflows section, click the workflow you want to rename to open the workflow canvas.

💡

Quick access

You can also access workflows from the Runs section by clicking on any run, then selecting View Workflow.

Step 2: Click the workflow name

At the top of the workflow canvas, click on the current workflow name (e.g., "Untitled Workflow").

The name becomes editable with a text input field.

Step 3: Enter new name

Type your new workflow name in the text field.

Naming best practices:

• ✅ Use descriptive names that indicate purpose: PCB-Component-Detection
• ✅ Include key details: model architecture, task, or version: Vehicle-VQA-Qwen2.5-7B-v2
• ✅ Use consistent naming conventions across workflows: {Domain}-{Task}-{Model}-{Version}
• ❌ Avoid generic names: Workflow 1, Test, My Model
• ❌ Don't use special characters that may cause issues: / \ : * ? " < > |

Step 4: Save the new name

Press Enter or click outside the text field to save your changes.

The workflow name updates immediately across:

• Workflow canvas header
• Workflows list
• Associated training runs
• Run history and logs

✅

Changes saved automatically

Your new workflow name is saved instantly. No additional confirmation required.

Workflow name requirements

Naming strategies

Common questions

Best practices

Use consistent naming conventions

Establish naming standards for your team or project:

Standard convention example:
{Project}-{Task}-{Model}-{TrainingMode}-v{Version}

PCB-Grounding-Qwen2.5-7B-LoRA-v1
PCB-Grounding-Qwen2.5-7B-LoRA-v2
PCB-VQA-NVLM-8B-FullFT-v1

Include key identifying information

Make it clear what makes this workflow unique:

• Task type: Phrase grounding vs. VQA
• Domain or use case: PCB inspection, defect detection, etc.
• Model architecture: Qwen2.5, NVLM, InternVL2.5
• Key parameters: Training mode, learning rate, batch size (for experiments)
• Version or iteration: v1, v2, v3 or date stamps

Rename before starting runs

Rename workflows from "Untitled Workflow" before starting your first training run for clearer run history:

Poor workflow:

Untitled Workflow → Run 1, Run 2, Run 3

Good workflow:

PCB-Detection-Qwen2.5-7B-v1 → Run 1, Run 2, Run 3

This makes it immediately clear which configuration produced each run.

Update names when configuration changes significantly

If you modify a workflow substantially (e.g., changing model architecture or completely rewriting the system prompt), consider updating the name or incrementing the version:

Example evolution:

PCB-Detection-v1 (Original)
  ↓ (Changed model from Qwen 2B to 7B)
PCB-Detection-Qwen7B-v2
  ↓ (Switched to full fine-tuning)
PCB-Detection-Qwen7B-FullFT-v3

Related resources

• Create a Workflow — Set up new training workflows
• Delete a Workflow — Remove workflows you no longer need
• Manage Workflows — Overview of workflow management options
• Start a Training Run — Launch training with your workflow