simonw

# /// script
# dependencies = [
#   "atproto"
# ]
# ///
from atproto import Client
import getpass
import time

sayakpaul

import torch
from diffusers import FluxTransformer2DModel
import torch.utils.benchmark as benchmark
from torchao.quantization import quantize_, int8_weight_only
from torchao.utils import unwrap_tensor_subclass
import torch._inductor

torch._inductor.config.mixed_mm_choice = "triton"

linoytsaban

# download FluxCFGPipline
!wget https://raw.githubusercontent.com/linoytsaban/diffusers/refs/heads/dreambooth-lora-flux-exploration/examples/community/pipeline_flux_with_cfg.py

# load pipeline
import diffusers
import torch
from pipeline_flux_with_cfg import FluxCFGPipeline

pipe = FluxCFGPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", 
                                    torch_dtype=torch.bfloat16)

sayakpaul

Running Flux under limited resources with Diffusers

Flux: https://blackforestlabs.ai/announcing-black-forest-labs/

• Run Flux with quantization by AmericanPresidentJimmyCarter.
• Run Flux on a 24GB 4090 by decoupling the different stages of the pipeline
• Running with torchao
• Running with NF4

The first resource even allows you to run the pipeline under 16GBs of GPU VRAM.

pcuenca

import argparse
import numpy as np
import torch
import torch.nn as nn
import coremltools as ct
from transformers import AutoTokenizer, AutoModelForCausalLM

# When using float16, all predicted logits are 0. To be debugged.
compute_precision = ct.precision.FLOAT32
compute_units = ct.ComputeUnit.CPU_ONLY

Artefact2

Which GGUF is right for me? (Opinionated)

Good question! I am collecting human data on how quantization affects outputs. See here for more information: ggml-org/llama.cpp#5962

In the meantime, use the largest that fully fits in your GPU. If you can comfortably fit Q4_K_S, try using a model with more parameters.

llama.cpp feature matrix

See the wiki upstream: https://github.com/ggerganov/llama.cpp/wiki/Feature-matrix

takanotaiga

# Copyright 2023 Taiga Takano
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http:#www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,

eavae

from pathlib import Path
from diffusers import StableDiffusionXLPipeline
import torch
from safetensors.torch import save_file

# text_encoder.text_model.encoder.layers.0.self_attn.k_proj.lora_linear_layer.down.weight
# lora_te_text_model_encoder_layers_0_self_attn_k_proj.lora_down.weight
# 1. text_encoder -> lora_te, text_encoder_2 -> lora_te2
# 2. map
# 3. .weight -> 2 .alpha -> 1 and replace . -> _

ypeleg

import json
import pickle
import struct
import zipfile
import numpy as np
from sentencepiece import SentencePieceProcessor

def rms_norm(x): return (x / np.sqrt(np.square(x).mean(-1, keepdims=True) + 1e-6))
def softmax(x): return (np.exp(x - np.max(x, axis=-1, keepdims=True))) / np.sum((np.exp(x - np.max(x, axis=-1, keepdims=True))), axis=-1, keepdims = True)

laksjdjf

'''
It supports only SD-v2 models.

usage:
python simo2kohya.py --unet <simo's unet weight path> --text <simo's text_encoder weight path> --save_to <save path>
(--text is optional)

This code may no longer be available due to updates from both @kohya-ss and @cloneofsimo.
'''