Update project links and hardware note

README.md

@@ -14,7 +14,9 @@ tags: [kandinsky, image-to-video, physical-ai, world-model]
 **Image-to-Video generation for Physical AI: autonomous driving · robotics · general physics**
 
 [🤗 Checkpoints collection](https://huggingface.co/collections/kandinskylab/kandinsky-wm-10-6a67455245052645fcfbd5f3) ·
-[🧩 Base model (Kandinsky 5.0)](https://github.com/kandinskylab/kandinsky-5)
+[🧩 Base model (Kandinsky 5.0)](https://github.com/kandinskylab/kandinsky-5) ·
+[💻 GitHub](https://github.com/kandinskylab/kandinsky-wm) ·
+[📝 Habr article](https://habr.com/ru/companies/sberbank/articles/1054966/)
 
 </div>
 
@@ -220,8 +222,7 @@ frames = pipe(
 export_to_video(frames, "robotics_out.mp4", fps=24, quality=9)
 ```
 
-Script: [`examples/run_i2v_diffusers.py`](https://github.com/kandinskylab/kandinsky-wm/blob/main/examples/run_i2v_diffusers.py). Lite (2B, ~21 GB bf16)
-fits a single 80 GB GPU with no offload; ~4 min for 50 steps on one H100.
+Script: [`examples/run_i2v_diffusers.py`](https://github.com/kandinskylab/kandinsky-wm/blob/main/examples/run_i2v_diffusers.py).
 
 
 ## Acknowledgements