Home
Creators
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

HiDream 1-Click Installer with an amazing Gradio App for Windows, RunPod and Massed Compute - NF4 Quantized Version (Patreon)

Published:
2025-04-13 22:30:52
Edited:
2025-10-16 10:46:23
Imported:
2026-06

Downloads

Content

1-Click to install and use HiDream with NF4 Quantization on Windows, RunPod and Massed Compute with pre-compiled wheels and model downlad

Patreon exclusive posts index to find our scripts easily, Patreon scripts updates history to see which updates arrived to which scripts and amazing Patreon special generative scripts list that you can use in any of your task.

Join discord to get help, chat, discuss and also tell me your discord username to get your special rank : SECourses Discord

Please also Star, Watch and Fork our Stable Diffusion & Generative AI  GitHub repository and join our Reddit subreddit and follow me on LinkedIn (my real profile)

=======

Latest installer : HiDream_v3.zip

The app will get more features hopefully soon - I had to rush publish since requested

It downloads NF4 quantized versions into models folder

Since NF4 quantized they use much lesser VRAM than original model

First run will download Fast model and Dev and Full models will be downloaded only if you use

All installation and model downloading is 100% automatic and works perfect

Supports RTX 5000 series and installs pre-compiled Flash Attention 2.8.3 (both Windows and Linux)

Ignore below errors on Windows CMD

CUDA extension not installed.

16 October 2025 Update

Installers updated to latest libraries such as CUDA 12.9, Torch 2.8, Flash Attention 2.8.3

Working amazing again, tested and verified

Ignore this error on CMD : CUDA extension not installed.

14 April 2025 Update

App updated to V3

Interface improved and custom any resolution usage feature added

int8 LLaMA implemented so it can produce better quality results - uses more VRAM

Windows Requirements

Python 3.10.11, FFmpeg, CUDA 12.9 or above, cuDNN 9.4 or above, C++ tools, MSVC and Git

If it doesn't work make sure to below tutorial and install everything exactly as shown in this below tutorial

https://youtu.be/DrhUHnYfwC0

Massed Compute (Recommend Cloud) :

Please register via this link : https://vm.massedcompute.com/signup?linkId=lp_034338&sourceId=secourses&tenantId=massed-compute

Use our coupon SECourses

Our coupon works on all GPUs now

H100 has amazing price and speed but you can use like RTX A6000 ADA as well

Full details here : https://www.patreon.com/posts/26671823

Then select our image SECourses from Creator dropdown

Then follow Massed_Compute_Instructions_READ.txt

Same as my any other Massed Compute installer script

Example tutorial for learn how to install and use Massed Compute

(Starts at 12:58) : https://youtu.be/KW-MHmoNcqo?si=G1WbG-Qw4ujWvOtG&t=778

RunPod (Cloud):

Please register via this link : https://get.runpod.io/955rkuppqv4h

Then follow Runpod_Instructions_READ.txt

Same as my any other RunPod installer script

Use the template written in Runpod_Instructions_READ.txt file

Example tutorial for learn how to install and use RunPod

(starts at 22:03) : https://youtu.be/KW-MHmoNcqo?si=QN8X8Sjn13ZYu-EU&t=1323

Files

Comments

Neil Rhodes

installing will check it out and see what the "noise" is all about :) thanks!

Dazzastrous

Yess was waiting for this the Doc is back in the game :) cheers Doc

cool1

Can the app please mention the VRAM requirements of each setting please (maybe near those options). eg. I tried the dev option and I think the cmd prompt said something about 27 GB VRAM, when my card only has 24 GB. Also could we have a stop/cancel option to stop image generation/multi image generation eg. if it's taking too long or we want to change one of the settings please?

Alexandr Mishuk

I don't like the Hi Dream. It generates for a long time. It's very similar to Flux. I'll stay with Flux for now. And maybe REVE. Only REVE is not convenient that there is no batch and all generation on the site. Thanks for giving me the opportunity to test it locally :)

cool1

Though unlike Flux dev, all the models of HiDream are a better license that allows commercial use if you wanted. On Flux only the Schnell option allows commercial use I think. Though we can also have Loras with Flux which we can't currently with HiDream. HiDream does seem a lot slower than Flux (on a 24 GB 3090 GPU) and I think there's problems with generating text in it but it probably depends on the model. Maybe the full Hidream is better. It's a shame HiDream needs so much VRAM and is slower than the other (though there's the fast option but I'm not sure it's as fast as Flux/schell and the fast won't be as good qualtiy).

Richard Young

can we select the token size ? -> WARNING:hdi1.pipelines.hidream_image.pipeline_hidream_image:The following part of your input was truncated because CLIP can only handle sequences up to 218 tokens: [''] WARNING:hdi1.pipelines.hidream_image.pipeline_hidream_image:The following part of your input was truncated because `max_sequence_length`

Rudbby

The has to be the most worth it Patreon I have joined! Thanks for this, was gonna ask for it!

Stevie Tee

Hi I'm receiving the following errors, should I pip install the missing extensions inside the activated venv ? Here: Hugging Face cache folders set to: E:\HiDream_v1\HiDream-I1-nf4\models W0415 13:06:54.390000 24812 venv\Lib\site-packages\torch\distributed\elastic\multiprocessing\redirects.py:29] NOTE: Redirects are currently not supported in Windows or MacOs. [2025-04-15 13:06:56,735] [INFO] [real_accelerator.py:222:get_accelerator] Setting ds_accelerator to cuda (auto detect) test.c LINK : fatal error LNK1181: cannot open input file 'aio.lib' test.c LINK : fatal error LNK1181: cannot open input file 'cufile.lib' Loading default model (fast)... ✅ Tokenizer loaded! (used 0.00 MB VRAM) WARNING:auto_gptq.nn_modules.qlinear.qlinear_cuda:CUDA extension not installed. WARNING:auto_gptq.nn_modules.qlinear.qlinear_cuda_old:CUDA extension not installed.

Gerenier

Can you add "John6666/Llama-3.1-8B-Lexi-Uncensored-V2-nf4" as optional encoder. This would be nice.

Mikel

Thank you for this! I've tried other install method and this is the only one that works for me (and my 4090.) I'm still on the fence about the perceived VRAM resource needed for HiDream, but honestly I feel like we need something new and fresh as Flux has been kicking around for a while. That said, this is a hobby for me and I enjoy running things local and it feels like I'm starting to be priced out.

DanO..

My friend has been getting some great results with HiDream and i2i in ComfyUI (Which I hate. XD). Can you add an i2i capability?

Pupattolino 75

Hello is it normal that in the terminal I receive this message? WARNING:auto_gptq.nn_modules.qlinear.qlinear_cuda:CUDA extension not installed. WARNING:auto_gptq.nn_modules.qlinear.qlinear_cuda_old:CUDA extension not installed.

Pupattolino 75

What I noticed that Fast model works fine but DEV is very slow. I presume because I have only a 16gb GPU (4090 lenovo laptop).? And with dev nf4 are necessary 21Gb vram? thanks in adv.

cool1

I clicked on "update_windows" to update to a newer version. I first (after that update) tried with llama model Int4 and it generated an image. When I changed the Llama model to int8 (I think) I got this error: File "G:\HiDream-AI-image-generator-app\HiDream-I1-nf4\venv\lib\site-packages\transformers\modeling_utils.py", line 1100, in _get_resolved_checkpoint_files raise EnvironmentError( OSError: clowman/Llama-3.1-8B-Instruct-GPTQ-Int8 does not appear to have a file named pytorch_model.bin, model.safetensors, tf_model.h5, model.ckpt or flax_model.msgpack. Should I have downloaded a new .zip of hidream or should just running "windows_update.bat" have updating everything needed?

cool1

Thanks. I'll look into that. I thought your HiDream GUI made things simpler though and I already downloaded that. With the SwarmUI it seems like a lot of other things need to be done (because it handles multiple models and has other different settings that need to be set up I think).

Jan Zhor

not working on my 5090, when i run I get: WARNING:bitsandbytes.cextension:The installed version of bitsandbytes was compiled without GPU support. 8-bit optimizers and GPU quantization are unavailable. Traceback (most recent call last): File "c:\HiDream_v2\HiDream-I1-nf4\hdi1\models\attention_processor.py", line 6, in from flash_attn_interface import flash_attn_func ModuleNotFoundError: No module named 'flash_attn_interface' During handling of the above exception, another exception occurred: Traceback (most recent call last): File "C:\Python3_10_11\lib\runpy.py", line 187, in _run_module_as_main mod_name, mod_spec, code = _get_module_details(mod_name, _Error) File "C:\Python3_10_11\lib\runpy.py", line 110, in _get_module_details __import__(pkg_name) File "c:\HiDream_v2\HiDream-I1-nf4\hdi1\__init__.py", line 1, in from .models.transformers.transformer_hidream_image import HiDreamImageTransformer2DModel File "c:\HiDream_v2\HiDream-I1-nf4\hdi1\models\transformers\transformer_hidream_image.py", line 16, in from ..attention_processor import HiDreamAttnProcessor_flashattn File "c:\HiDream_v2\HiDream-I1-nf4\hdi1\models\attention_processor.py", line 9, in from flash_attn import flash_attn_func File "c:\HiDream_v2\HiDream-I1-nf4\venv\lib\site-packages\flash_attn\__init__.py", line 3, in from flash_attn.flash_attn_interface import ( File "c:\HiDream_v2\HiDream-I1-nf4\venv\lib\site-packages\flash_attn\flash_attn_interface.py", line 15, in import flash_attn_2_cuda as flash_attn_gpu ImportError: DLL load failed while importing flash_attn_2_cuda: The specified module could not be found. ----- during install I get: Looking in indexes: https://download.pytorch.org/whl/nightly/cu128 ERROR: Could not find a version that satisfies the requirement torch==2.7.0.dev20250311 (from versions: 2.7.0.dev20250310+cu128, 2.8.0.dev20250427+cu128, 2.8.0.dev20250428+cu128, 2.8.0.dev20250429+cu128, 2.8.0.dev20250430+cu128, 2.8.0.dev20250501+cu128, 2.8.0.dev20250502+cu128, 2.8.0.dev20250503+cu128, 2.8.0.dev20250504+cu128, 2.8.0.dev20250505+cu128, 2.8.0.dev20250506+cu128, 2.8.0.dev20250507+cu128, 2.8.0.dev20250508+cu128, 2.8.0.dev20250509+cu128, 2.8.0.dev20250510+cu128, 2.8.0.dev20250511+cu128, 2.8.0.dev20250512+cu128, 2.8.0.dev20250513+cu128, 2.8.0.dev20250514+cu128, 2.8.0.dev20250515+cu128, 2.8.0.dev20250516+cu128, 2.8.0.dev20250517+cu128, 2.8.0.dev20250518+cu128, 2.8.0.dev20250519+cu128, 2.8.0.dev20250520+cu128, 2.8.0.dev20250521+cu128, 2.8.0.dev20250522+cu128, 2.8.0.dev20250523+cu128, 2.8.0.dev20250524+cu128, 2.8.0.dev20250525+cu128, 2.8.0.dev20250526+cu128, 2.8.0.dev20250527+cu128, 2.8.0.dev20250528+cu128, 2.8.0.dev20250529+cu128, 2.8.0.dev20250530+cu128, 2.8.0.dev20250531+cu128, 2.8.0.dev20250601+cu128, 2.8.0.dev20250602+cu128, 2.8.0.dev20250603+cu128, 2.8.0.dev20250604+cu128, 2.8.0.dev20250605+cu128, 2.8.0.dev20250606+cu128, 2.8.0.dev20250607+cu128, 2.8.0.dev20250608+cu128, 2.8.0.dev20250609+cu128, 2.8.0.dev20250610+cu128, 2.8.0.dev20250611+cu128, 2.8.0.dev20250612+cu128, 2.8.0.dev20250613+cu128, 2.8.0.dev20250614+cu128, 2.8.0.dev20250615+cu128, 2.8.0.dev20250616+cu128, 2.8.0.dev20250617+cu128, 2.8.0.dev20250618+cu128, 2.8.0.dev20250619+cu128, 2.8.0.dev20250620+cu128, 2.8.0.dev20250621+cu128, 2.8.0.dev20250622+cu128, 2.8.0.dev20250623+cu128, 2.8.0.dev20250624+cu128, 2.8.0.dev20250625+cu128) ERROR: No matching distribution found for torch==2.7.0.dev20250311 Collecting flash-attn==2.7.4.post1 Downloading https://huggingface.co/MonsterMMORPG/SECourses_Premium_Flash_Attention/resolve/main/flash_attn-2.7.4.post1-cp310-cp310-win_amd64.whl (138.2 MB) ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 138.2/138.2 MB 12.0 MB/s eta 0:00:00 Requirement already satisfied: torch in c:\hidream_v2\hidream-i1-nf4\venv\lib\site-packages (from flash-attn==2.7.4.post1) (2.7.1) any clue?

Loyd Keudem

Edit this file and set set HF_HUB_ENABLE_HF_TRANSFER=0 if you get model download errors of HF_TRANSFER Hugging Face cache folders set to: E:\AI_Lab\HiDream_v2\HiDream-I1-nf4\models Skipping import of cpp extensions due to incompatible torch version 2.7.0+cu128 for torchao version 0.14.0 Please see GitHub issue #2919 for more info W1015 18:42:31.903212 24336 venv\Lib\site-packages\torch\distributed\elastic\multiprocessing\redirects.py:29] NOTE: Redirects are currently not supported in Windows or MacOs. Loading default model (fast)... ✅ Tokenizer loaded! (used 0.00 MB VRAM) CUDA extension not installed. CUDA extension not installed. I get two errors: Skipping import of cpp extensions due to incompatible torch version 2.7.0+cu128 for torchao version 0.14.0 Please see GitHub issue #2919 for more info And the most important error: CUDA extension not installed. CUDA extension not installed Help fix I have a 5070Ti as GPU