INSANE Image Understanding/Thinking | Deepseek Janus Pro | Complete Crash Course

[@TroubleChute]

Looking for Deepseek R1? See the guide here: kzbin.info/www/bejne/qWLch4Z3n753aq8 (Also free, and local/offline)

[@Poxeld]

do you have any possible solution for the cod "can't join because you are on a different version" bug?

[@vincentnestler1805]

Thanks!

[@pedrolinares6309]

Great video, when Deepseek went local I did not know where to find the right steps to run it locally with UI. This video and the one of the chatbox save so much time! Thanks!

[@JohnSmith-iz9qn]

Thanks for a great video! Just wanted to say I encountered a few more issues. Needed to: 1. Install Python v3.12.8 (the newest v3.13 doesn't yet have support for PyO3 library for rust, which was needed later) 2. Modify my Visual Studio installation (add MSVC v142 - VS 2019 C++ build tools) 3. Install CMake (v3.31.5) 4. Install Rust compiler (v1.84.0) 5. Reinstall PyTorch (the default one didn't have CUDA support for my RTX 4070 Ti Super) But at the end, it works. I'm using the 7b model and it usually takes ~10s to generate or describe an image, although it depends on the prompt (for simple prompts it returns result in ~1s).

[@TheJunky228]

very helpful and uploaded 24 minutes before I looked it up! thank you for this guide

[@paultsjan6047]

Janus Pro is an advanced multimodal AI model developed by DeepSeek. It is designed to handle both multimodal understanding and text-to-image generation tasks effectively. The model can generate high-quality images from text descriptions and also understand visuals, which is becoming increasingly important in modern AI applications. Janus Pro-7B outperforms models like OpenAI's DALL-E 3 and Stable Diffusion on benchmarks such as GenEval and DPG-Bench, thanks to its improved training processes, data quality, and model size. Janus Pro represents a significant advancement in the field of multimodal AI, offering a versatile and powerful tool for developers and researchers working with text and image data.

[@Fx_Explains]

Nice, I hope it helps me on my youtube channels, I don't need complex images, just simple ones.

[@clouds2593]

Good video.

[@DHell-n8i]

Im getting this problem FileNotFoundError: [WinError 2] The system cannot find the file specified [end of output] note: This error originates from a subprocess, and is likely not a problem with pip. error: subprocess-exited-with-error × Getting requirements to build wheel did not run successfully. │ exit code: 1 ╰─> See above for output.

[@HogbergPhotography]

The Huggingface link does not work.

[@Whitect4v]

FileNotFoundError: [WinError 2] The system cannot find the file specified [end of output] note: This error originates from a subprocess, and is likely not a problem with pip. error: subprocess-exited-with-error × Getting requirements to build wheel did not run successfully. │ exit code: 1 ╰─> See above for output.

[@prodginm]

how do you get the .git folder? 7:50 i have hidden folders on and still cant see it

[@TroubleChute]

Only if you use the git clone command. Downloading the zip doesn't include this. It's not important to the program working

[@prodginm]

@@TroubleChute thank you

[@Phoenix-tc6dp]

its so weird putting a face to a voice i always thought you where some bald guy with a beard lol

[@av7062]

Yeah, instead he's some guy with hair and a beard! 😮

[@clouds2593]

By the looks of it, you'll be right soon.

[@boomvideo1776]

Bro how to reach to you personally, I have some doubts about installation

[@aurav4643]

Is it viable to use it via ZLUDA for an AMD GPU ? I do think it'd be wiser for me to use WSL and use it through a Linux for the ROCm PyTorch

[@JuanDiegoPinillos]

Apparently yes it is, in fact, AMD gpus are faster using DeepSeek than Nvidia's cards.

[@trickveira]

I managed to run it on the RX Vega 56 GPU through Linux using ROCm (Nobara OS / fedora 41), but it is extremely slow, slower than FLUX and SD 3.5. I will wait for some update or detailed configuration...

[@vinm300]

LOL - very entertaining

[@wassupdoc7742]

make a tutorial about how to create an AI avatar like yours

[@URBANSEE]

que mal explicado el video de instalación 1/10 (haces un montón de pasos no necesarios)

[@JuanDiegoPinillos]

In the Breaking bad image the LLM wasn't wrong, it was giving you an answer from ITS PERSPECTIVE! its left and right are the opposite as yours. creepy but fascinating stuff.

[@SoulSeeker]

The double you, was not a mistake, the hand written note has two of them next to each other.

[@TheJunky228]

look more closely, it was a triple you which was the mistake

[@sixthsick]

Your voice does not match with your face man. IDK why but it doesnt.

[@bluearrow9331]

it do not work and I'm mad because you just explain everything so fast like every one who watching your video is an expert in running AI so here is the error and please tell us how to do everything error:- 'conda' is not recognized as an internal or external command, operable program or batch file.

[@Atttalie]

Un-subscribing for support to AI.

[@febsho9613]

Oh no

[@shivpratapsingh2881]

You’ll be missed…not. 😢