sayakpaul

When loading the LoRA params (that were obtained on a quantized base model) and merging them into the base model, it is recommended to first dequantize the base model, merge the LoRA params into it, and then quantize the model again. This is because merging into 4bit quantized models can lead to some rounding errors. Below, we provide an end-to-end example:

1. First, load the original model and merge the LoRA params into it:

from diffusers import FluxPipeline 
import torch 

ckpt_id = "black-forest-labs/FLUX.1-dev"
pipeline = FluxPipeline.from_pretrained(

cubiq

{
  "last_node_id": 469,
  "last_link_id": 1401,
  "nodes": [
    {
      "id": 16,
      "type": "KSamplerSelect",
      "pos": [
        -280,
        20

lucataco

# conda create -n flux python=3.11
# conda activate flux
# pip install torch==2.3.1
# pip install diffusers==0.30.0 transformers==4.43.3
# pip install sentencepiece==0.2.0 accelerate==0.33.0 protobuf==5.27.3

import torch
from diffusers import  FluxPipeline
import diffusers

awni

lucataco

# Install Ubuntu 22.04
sudo apt-get update
sudo apt-get upgrade -y

# Install ssh, curl, git, htop
sudo apt install openssh-server 
sudo apt install curl git htop zstd

# Install CUDA toolkit 12.4 drivers
https://developer.nvidia.com/cuda-downloads