Update README.md
Browse files
README.md
CHANGED
|
@@ -9,18 +9,10 @@ pinned: false
|
|
| 9 |
|
| 10 |
# Vierla
|
| 11 |
|
| 12 |
-
|
| 13 |
|
| 14 |
-
We
|
| 15 |
|
| 16 |
-
|
| 17 |
-
- **Efficient inference:** GPU memory tiering (VRAM → system RAM → disk), CUDA VMM, 70B+ models on consumer GPUs
|
| 18 |
-
- **Training efficiency:** optimizers and speedruns (Muon, modded-nanogpt)
|
| 19 |
-
- **3D & vision:** Gaussian splatting, optimal-transport morphs, vision foundation models
|
| 20 |
-
- **Agent infrastructure:** MCP tooling, local control planes for coding agents
|
| 21 |
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
Toronto, Canada · Google for Startups Cloud Program
|
| 25 |
-
|
| 26 |
-
🌐 [vierla.com](https://vierla.com) · [GitHub](https://github.com/vierla) · [hello@vierla.com](mailto:hello@vierla.com)
|
|
|
|
| 9 |
|
| 10 |
# Vierla
|
| 11 |
|
| 12 |
+
AI research lab in Toronto. Models, datasets and demos from our work land here.
|
| 13 |
|
| 14 |
+
We work on memory, representation and world models, and on the engineering that makes big models cheap to run: GPU memory tiering for 70B+ models on consumer GPUs, fast training (Muon, modded-nanogpt), Gaussian splatting.
|
| 15 |
|
| 16 |
+
Nothing public yet.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
|
| 18 |
+
[vierla.com](https://vierla.com) · [GitHub](https://github.com/Vierla)
|
|
|
|
|
|
|
|
|
|
|
|