Update README.md

README.md

@@ -14,12 +14,14 @@ We introduce **GroveMoE**, a new sparse architecture using **adjugate experts**
 - **Sparse Activation**: 33 B params total, only **3.14–3.28 B** active per token.
 - **Traning**: Mid-training + SFT, up-cycled from Qwen3-30B-A3B-Base; preserves prior knowledge while adding new capabilities.
 
-## Model Lists
-| GroveMoE Series | Download  
-|---|---
-GroveMoE-Base   | 🤗 [HuggingFace](https://huggingface.co/inclusionAI/GroveMoE-Base)
-GroveMoE-Inst  | 🤗 [HuggingFace](https://huggingface.co/inclusionAI/GroveMoE-Inst)
+## Model Downloads
 
+<div align="center">
+|     **Model**      | **#Total Params** | **#Activated Params** | **Download** |
+| :----------------: | :---------------: | :-------------------: | :----------: |
+| GroveMoE-Base |       33B       |         3.14~3.28B         | [🤗 HuggingFace](https://huggingface.co/inclusionAI/Base) |
+| GroveMoE-Inst |       3B       |         3.14~3.28B         | [🤗 HuggingFace](https://huggingface.co/inclusionAI/GroveMoE-Inst) |
+</div>
 
 ## Performance
 
@@ -32,7 +34,7 @@ GroveMoE-Inst  | 🤗 [HuggingFace](https://huggingface.co/inclusionAI/GroveMoE-
 |Mistral-Small-3.2| 24B | 68.1 | 37.5 | 59.9 | 61.9 | 33.4 | 28.1 | 69.5 | 32.2 |
 |GroveMoE-Inst|3.14~3.28B | <font color=#FBD98D>**72.8**</font> | <font color=#FBD98D>**47.7**</font> | <font color=#FBD98D>**61.3**</font> |<font color=#FBD98D>**71.2**</font> |<font color=#FBD98D>**43.5**</font> | <font color=#FBD98D>**44.4**</font> |<font color=#FBD98D>**74.5**</font> | <font color=#FBD98D>**34.6**</font> |
 
-We bold the top1 scores separately for all models.
+We bold the top1 scores separately for all models. More details will be reported in our [technical report](https://arxiv.org/abs/2508.07785).
 
 ## Usage
 Below, there are some code snippets on how to get quickly started with running the model. First, install the Transformers library.