Emil Bogomolov zetyquickly

🎰

<-- generated prior COVID

zetyquickly / stablediffusionwalk.py

Created April 12, 2025 23:27 — forked from karpathy/stablediffusionwalk.py

hacky stablediffusion code for generating videos

	"""
	stable diffusion dreaming
	creates hypnotic moving videos by smoothly walking randomly through the sample space

	example way to run this script:

	$ python stablediffusionwalk.py --prompt "blueberry spaghetti" --name blueberry

	to stitch together the images, e.g.:
	$ ffmpeg -r 10 -f image2 -s 512x512 -i blueberry/frame%06d.jpg -vcodec libx264 -crf 10 -pix_fmt yuv420p blueberry.mp4

zetyquickly / stablediffusionwalk.py

Created April 12, 2025 23:27 — forked from karpathy/stablediffusionwalk.py

hacky stablediffusion code for generating videos

	"""
	stable diffusion dreaming
	creates hypnotic moving videos by smoothly walking randomly through the sample space

	example way to run this script:

	$ python stablediffusionwalk.py --prompt "blueberry spaghetti" --name blueberry

	to stitch together the images, e.g.:
	$ ffmpeg -r 10 -f image2 -s 512x512 -i blueberry/frame%06d.jpg -vcodec libx264 -crf 10 -pix_fmt yuv420p blueberry.mp4

zetyquickly / view_contig.py

Last active March 15, 2022 03:36

View Contiguous

	import torch

	# Tensor of interest
	tens_A = torch.rand((2,3,4)) # 3-dimensional tensor of shape (2,3,4)

	def test_view(tensor, sizes):
	try:
	tensor.view(*sizes)
	except Exception as e:
	print(e)

zetyquickly / contig_and_again.py

Last active March 14, 2022 04:15

Non-contiguous and restoration

	import torch

	# Tensor of interest
	tens_A = torch.rand((2,3,4)) # 3-dimensional tensor of shape (2,3,4)
	# Transposed tensor
	transp_A = tens_A.t()
	# Check contiguity
	print(transp_A.is_contiguous()) # output: False
	# Make contiguous
	tens_B = transp_A.contiguous() # triggers copying because wasn't contiguous

zetyquickly / stride_product_tail.py

Created March 14, 2022 03:52

Strides are tail products

	import torch

	# Tensor of interest
	tens_A = torch.rand((2,3,4)) # 3-dimensional tensor of shape (2,3,4)
	sizes = tens_A.shape

	print(tens_A.is_contiguous()) # outputs: True
	print(tens_A.stride()) # outputs: (12,4,1)
	print(tens_A.stride() == (sizes[1]*sizes[2], sizes[2], 1)) # outputs: True

zetyquickly / stride_offset_jump.py

Last active March 14, 2022 00:19

Tensor stride is a jump within underlying data

	import torch

	# Tensor of interest
	tens_A = torch.rand((2,3,4)) # 3-dimensional tensor of shape (2,3,4)

	# Start index
	x = 0
	y = 1
	z = 3

zetyquickly / und_data_resh.py

Last active March 13, 2022 22:48

Tensor reshape 2x4 to 2x2x2

	import torch

	data = [1,2,3,4,5,6,7,8]

	# Original tensor of shape 2x4
	tens_A = torch.tensor(data).reshape(shape=(2,4))
	# Reshaped from 2x4 to 2x2x2 (preserving number of elements)
	tens_B = tens_A.reshape(shape=(2,2,2))

zetyquickly / und_data_init.py

Last active March 13, 2022 22:47

Underlying data reshape

	import torch

	# underlying data
	data = [1,2,3,4,5,6,7,8] # has 8 elements

	# two ways to store identical data
	tens_A = torch.tensor(data).reshape(shape=(2,4)) # 2-dimensional tensor of shape (2,4)
	tens_B = torch.tensor(data).reshape(shape=(2,2,2)) # 3-dimensional tensor of shape (2,2,2)

zetyquickly / baseline.py

Last active January 26, 2022 08:34

Codenrock New Year Battle Baseline Model

	import torch
	from torch import nn
	import torchvision
	from torchvision import transforms
	from torch.utils.data import Dataset

	from PIL import Image
	import os

	from sklearn.model_selection import train_test_split

zetyquickly / full_pool.py

Last active July 29, 2021 23:52

	import itertools
	import time

	import numpy as np
	from numpy.random import default_rng

	import trimesh
	import igl
	from tqdm import tqdm