This repository implements a one-step generative Flow Matching model based on MeanFlow.
Flowers:
1) a close up of a sword lily with orange flowers
2) a watercress in a flower pot
3) a yellow bougainvillea in the middle of the plant
4) a yellow columbine in the garden
5) a wallflower in the garden
6) a white silverbush in the middle of a flower
7) a blanket flower with a yellow and red flower
8) a small blue stemless gentian in the grass
9) a close up of a white petunia
10) a pink siam tulip with white flowers on a green background
11) a pink ruby-lipped cattleya with a white background
12) a close up of a red bee balm
COCO Dataset:
1) A brown teddy bear standing next to a toothbrush. | 1) A green umbrella sitting on top of a sandy beach.
2) Riding a motorcycle down a street that has no one ... | 2) A country charm type of kitchen is equipped with ...
3) The toilet is in a room with exposed pipes. | 3) A man riding a snowboard down the side of a ski slope.
4) A couple of men sitting at a table having dinner ... | 4) A man stands beside his black and red motorcycle...
5) A cat walking on the top of an open door. | 5) A calico cat standing on top of an upholstered chair.
6) The silhouette of people is seen against the inside of ... | 6) Many kites can be seen in the air through umbrellas.
7) A donut on a plate with a fork and knife. | 7) A baseball player holding a wooden bat standing.
8) A black dog running in a pen with a horse. | 8) A small kitchen inside of a dark office.
9) A bathroom with a sink, paper roll, toilet, towel... | 9) A herd of sheep standing in a snow-covered field.
10) A living room filled with furniture and a staircase. | 10) A table topped with glasses and eating utensils.
11) A game of tennis on a blue court with an audience. | 11) A little girl sitting at a table with lots of fruit ...
12) A person and a laptop in a room. | 12) There is a male skateboarder doing a trick.
from diffusers.models import AutoencoderKL
vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse").eval()
from transformers import AutoTokenizer, AutoModel
pre_tokenizer = AutoTokenizer.from_pretrained("intfloat/e5-base")
pre_model = AutoModel.from_pretrained("intfloat/e5-base")