Add project links to model card
Browse files
README.md
CHANGED
|
@@ -14,6 +14,8 @@ private: true
|
|
| 14 |
|
| 15 |
# Laguna Vision
|
| 16 |
|
|
|
|
|
|
|
| 17 |
Laguna Vision adds a visual input path to `poolside/Laguna-XS.2`. SigLIP encodes images, AnyRes tiling preserves screenshot/document detail, a resampler projector maps features into Laguna's embedding space, and LoRA adapters are trained with supervised visual-instruction data.
|
| 18 |
|
| 19 |
Method: **post-training multimodal adaptation via supervised fine-tuning**.
|
|
|
|
| 14 |
|
| 15 |
# Laguna Vision
|
| 16 |
|
| 17 |
+
[Open-source GitHub](https://github.com/aaronkazah/laguna-vision) 路 [Hugging Face model](https://huggingface.co/poolside-laguna-hackathon/laguna-vision)
|
| 18 |
+
|
| 19 |
Laguna Vision adds a visual input path to `poolside/Laguna-XS.2`. SigLIP encodes images, AnyRes tiling preserves screenshot/document detail, a resampler projector maps features into Laguna's embedding space, and LoRA adapters are trained with supervised visual-instruction data.
|
| 20 |
|
| 21 |
Method: **post-training multimodal adaptation via supervised fine-tuning**.
|