Difference between FlowMatchEulerDiscreteScheduler and UniPCMultistepScheduler, and compatibility for LoRA fine-tuned weights

[archernsy]

Hello! I have some questions regarding the differences between FlowMatchEulerDiscreteScheduler and UniPCMultistepScheduler:

（1）What are the key technical differences between these two schedulers?

（2）I used finetrainers to perform LoRA fine-tuning on Wan2.1 T2V 1.3B diffusers, which defaults to FlowMatchEulerDiscreteScheduler. However, my inference pipeline currently uses UniPCMultistepScheduler.

（3）Will using UniPCMultistepScheduler for inference with weights fine-tuned using FlowMatchEulerDiscreteScheduler affect the generation quality?

Thank you!

​​Additional context​​:

Model: Wan2.1 T2V 1.3B diffusers
Fine-tuning method: LoRA via finetrainers
Current setup: Fine-tuned with FlowMatchEulerDiscreteScheduler → Inference with UniPCMultistepScheduler