Update README.md
Browse files
README.md
CHANGED
@@ -15,8 +15,6 @@ It accepts both image and text inputs to generate text outputs.
|
|
15 |
|
16 |
Thanks to its lightweight design, it can be deployed on edge devices such as AI glasses and smartphones, offering low memory usage and high speed while maintaining strong performance on multimodal tasks. Some well-known small models include [PaliGemma 3B](https://huggingface.co/google/paligemma-3b-mix-448), [Moondream2](https://huggingface.co/vikhyatk/moondream2), [Qwen2-VL-2B](https://huggingface.co/Qwen/Qwen2-VL-2B), [InternVL2-2B](https://huggingface.co/OpenGVLab/InternVL2-2B), and [InternVL2_5-2B](https://huggingface.co/OpenGVLab/InternVL2_5-2B). Ivy-VL outperforms them on multiple benchmarks.
|
17 |
|
18 |
-
The model is built upon the `Qwen/Qwen2.5-3B-Instruct` language model, with [<code>google/siglip-so400m-patch14-384</code>](https://huggingface.co/google/siglip-so400m-patch14-384) serving as the vision encoder.
|
19 |
-
|
20 |
# Model Summary:
|
21 |
|
22 |
* Developed: Standford, CMU, AI Safeguard
|
@@ -28,7 +26,10 @@ The model is built upon the `Qwen/Qwen2.5-3B-Instruct` language model,
|
|
28 |
* License: Apache 2.0
|
29 |
|
30 |
* Architecture: Based on LLaVA-One-Vision
|
31 |
-
|
|
|
|
|
|
|
32 |
|
33 |
# Evaluation:
|
34 |
|
|
|
15 |
|
16 |
Thanks to its lightweight design, it can be deployed on edge devices such as AI glasses and smartphones, offering low memory usage and high speed while maintaining strong performance on multimodal tasks. Some well-known small models include [PaliGemma 3B](https://huggingface.co/google/paligemma-3b-mix-448), [Moondream2](https://huggingface.co/vikhyatk/moondream2), [Qwen2-VL-2B](https://huggingface.co/Qwen/Qwen2-VL-2B), [InternVL2-2B](https://huggingface.co/OpenGVLab/InternVL2-2B), and [InternVL2_5-2B](https://huggingface.co/OpenGVLab/InternVL2_5-2B). Ivy-VL outperforms them on multiple benchmarks.
|
17 |
|
|
|
|
|
18 |
# Model Summary:
|
19 |
|
20 |
* Developed: Standford, CMU, AI Safeguard
|
|
|
26 |
* License: Apache 2.0
|
27 |
|
28 |
* Architecture: Based on LLaVA-One-Vision
|
29 |
+
|
30 |
+
* LLM: Qwen/Qwen2.5-3B-Instruct
|
31 |
+
|
32 |
+
* Vision Encoder: google/siglip-so400m-patch14-384
|
33 |
|
34 |
# Evaluation:
|
35 |
|