Mochi-1 Preview LoRA Finetune

This is a LoRA fine-tune of the Mochi-1 preview model. The model was trained using custom training data.

Prompt: gingerbread man is dancing on a table. melty.

Usage

from diffusers import MochiPipeline
from diffusers.utils import export_to_video
import torch

pipe = MochiPipeline.from_pretrained("genmo/mochi-1-preview")
pipe.load_lora_weights("lucataco/mochi-lora-melty")
pipe.enable_model_cpu_offload()

video = pipe(
    prompt="your prompt here",
    guidance_scale=6.0,
    num_inference_steps=64,
    height=480,
    width=848,
    max_sequence_length=256,
).frames[0]

export_to_video(video, "output.mp4", fps=30)

Training details

Trained on Replicate using: lucataco/mochi-1-lora-trainer