It's all coming into focus.

🧵 10/10 💡 Ready to Get Started?. 📖 Read the full blog post and docs: ⭐ Star us on GitHub: . We can't wait to see what incredible things you create!.

🧵 9/10 🏢 Enterprise & Creator Use Cases. Companies: Build proprietary, on-brand image generators for internal teams. Artists: Create completely unique styles never seen before. Researchers: Push model boundaries while studying biases & capabilities. All private! 🛡️.

🧵 8/10 🎨 How The Simpsons LoRA Works. Trained on cartoon captions, it learned: ✨ Bright, vibrant colors 📐 Simple, clean outlines. 🎭 Classic flat shading. Result? ANY prompt becomes a fun, colorful Springfield scene! The training docs show you exactly how to do this.

🧵 7/10 📊 Training Results That'll Blow Your Mind. Prompt: "An astronaut floating in space".- Base model → Realistic astronaut 👨‍🚀.- Simpsons LoRA → Springfield cartoon style! 🟡. A tiny LoRA adapter completely transforms the style. The power is incredible!

🧵 6/10 🏋️ The Best Part: Custom Model Training. This is where Transformer Lab truly shines: .• Full job management system.• Real-time status tracking.• Workflow support (coming soon!). Train your own Diffusion LoRAs with enterprise-grade tools! 💪.

🧵 5/10 🔄 Img2Img Transformations. Feed an existing image + styling prompt = completely new artistic vision!. Take a regular photo and transform it into any style you can imagine. The before/after results are mind-blowing! 🤯

🧵 4/10 🎭 Native Inpainting Tool. Want to edit specific parts of an image? We built a paint tool RIGHT into Transformer Lab!. Example: Turned Audrey Hepburn's hair pink by simply masking the area. The precision is incredible! ✨

3/10 ✨ Image Generation Made Simple. Create stunning images with text prompts. This imag was generated using Flux [dev] on a dual RTX 3090 setup. The GUI makes it incredibly intuitive - just type and generate! 🎨

2/10 📦 Wide Model Support Out of the Box:. • Stable Diffusion (1.5, XL, 3) .• Flux (including [dev]).• Thousands of pre-trained LoRAs .• And many more models!. Everything runs privately - local or cloud. 🔒.

Our biggest update to date. Transformer Lab now supports Diffusion image generation and training!. ✨ Generate images .🎨 Inpaint & edit .🔄 Img2img transforms .🏋️ Train custom LoRAs. All running locally on YOUR hardware! Thread 👇.

Something very big is coming! 🖼️🪄.

Ready to try it? Check out our installation guide: and welcome to the Transformer Lab family, AMD users! 🚀 7/7.

While we had to adapt to work without bitsandbytes (which has open issues with ROCm), we're proud to have achieved full AMD support for both Linux and Windows users. 6/7.

We learned that version matching is critical—ROCm components, drivers, libraries, and PyTorch wheels all need to be on exactly the right versions or things break. 5/7.

After extensive troubleshooting and help from the ROCm community, we discovered and implemented a workaround for PyTorch detection issues on WSL—and we've automated it so our users don't have to worry about it. 4/7.

One of our biggest challenges? Getting ROCm (AMD's answer to CUDA) working properly on Windows via WSL. On paper, it should "just work" but reality had other plans! 3/7.

This was a (long) journey that involved building an AMD test rig from scratch and diving deep into the world of ROCm, drivers, and PyTorch on AMD. 2/7.

🎉 BIG NEWS: Transformer Lab now supports AMD GPUs on both Linux and Windows! You can now harness the power of your AMD hardware to run and train models with ease. #AMD 1/7.

Oh boy. we have some big product announcements coming. Hope to create a new build by end of week and will update you all here. Join our Discord to hear updates first.