# How to Create a Pipeline?

{% stepper %}
{% step %}

### Step 1

* **Select Base Model:**
  * Choose a foundational model from the Model Catalog (e.g., DeepSeek, Gemma, Llama, Qwen) or your Private Model repository. Models vary by size (from 0.5B to 72B parameters), type (LLM, VLM), and training stage (pre-trained or instruction-tuned).
* **Select Trainer:**
  * Pick a training method such as SFT (Supervised Fine-tuning), DPO (Direct Preference Optimization) or Pre-training, depending on your task and data type.
* **Select Volume:**
  * Select a storage type based on dataset size: Managed Volume for datasets under 20GiB or Dedicated Network Volume for larger datasets requiring manual provisioning.
    {% endstep %}

{% step %}

### Step 2

* **Select Dataset Format:**
  * Choose the appropriate format (Alpaca, ShareGPT, Corpus, etc.) based on your selected trainer.
* **Select Dataset:**
  * Upload training and evaluation datasets or connect to Data Hub. Ensure data quality, consistency, and correct formatting to optimize model performance
    {% endstep %}

{% step %}

### Step 3

* **Set up Hyperparameters:**
  * Configure training parameters like batch size, learning rate, epochs and optimizer settings. Advanced options are available for DPO, LoRA, and runtime optimization.
* **Set up Infrastructure:**
  * Supported both single-node and multi-node, with up to 16 nodes
    {% endstep %}

{% step %}

### Step 4

* **Finish:**
  * Name your pipeline and optionally add a description. You'll receive email notifications when the pipeline succeeds or fails
    {% endstep %}
    {% endstepper %}


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://ai-docs.fptcloud.com/fpt-ai-studio/services/model-fine-tuning/tutorials/how-to-create-a-pipeline.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.