[BUG] Parameters of `GeneratorStep` are ignored during `dry_run`

[sung1-kang]

Describe the bug

I found that the parameters assigned to GeneratorStep are ignored during dry_run, and only batch_size remains. This behavior seems incorrect, as it should preserve the existing parameters and adjust only the batch_size.

To reproduce

1. Configure a pipeline with a GeneratorStep that includes custom parameters.
2. Run the pipeline in dry_run mode.
3. Observe that the parameters assigned to the GeneratorStep are lost, and only batch_size is retained.

Expected behavior

During dry_run, the parameters assigned to GeneratorStep should be preserved, and only the batch_size should be adjusted.

Screenshots

No response

Environment

• Distilabel Version [1.5.3]:
• Python Version [3.11]:

Additional context

Actual Behavior

The parameters assigned to GeneratorStep are overwritten, leaving only the batch_size.

Code Snippet

The issue seems to originate from the following code snippet:

for step_name in self.dag:
    step = self.dag.get_step(step_name)[constants.STEP_ATTR_NAME]

    if step.is_generator:
        if not parameters:
            parameters = {}
        parameters[step_name] = {"batch_size": batch_size}

This code is located at: https://github.com/argilla-io/distilabel/blob/main/src/distilabel/pipeline/base.py#L438

Proposed Solution

Modify the code to preserve existing parameters and only update the batch_size. For example:

if not parameters:
    parameters = {}

for step_name in self.dag:
    step = self.dag.get_step(step_name)[constants.STEP_ATTR_NAME]

    if step.is_generator:
        if step_name not in parameters:
            parameters[step_name] = {}
        parameters[step_name]["batch_size"] = batch_size

This change ensures that existing parameters are not overwritten and only the batch_size is updated.