mirror of
https://github.com/OpenBMB/MiniCPM-V.git
synced 2026-02-04 09:49:20 +08:00
79
README.md
79
README.md
@@ -171,6 +171,85 @@ MiniCPM-V 4.5 can be easily used in various ways: (1) [llama.cpp](https://github
|
||||
</div>
|
||||
|
||||
|
||||
### Inference Efficiency
|
||||
|
||||
|
||||
**OpenCompass**
|
||||
<div align="left">
|
||||
<table style="margin: 0px auto;">
|
||||
<thead>
|
||||
<tr>
|
||||
<th align="left">Model</th>
|
||||
<th>Size</th>
|
||||
<th>Avg Score ↑</th>
|
||||
<th>Total Inference Time ↓</th>
|
||||
</tr>
|
||||
</thead>
|
||||
<tbody align="center">
|
||||
<tr>
|
||||
<td nowrap="nowrap" align="left">GLM-4.1V-9B-Thinking</td>
|
||||
<td>10.3B</td>
|
||||
<td>76.6</td>
|
||||
<td>17.5h</td>
|
||||
</tr>
|
||||
<tr>
|
||||
<td nowrap="nowrap" align="left">MiMo-VL-7B-RL</td>
|
||||
<td>8.3B</td>
|
||||
<td>76.4</td>
|
||||
<td>11h</td>
|
||||
</tr>
|
||||
<tr>
|
||||
<td nowrap="nowrap" align="left">MiniCPM-V-4_5</td>
|
||||
<td>8.7B</td>
|
||||
<td>77.0</td>
|
||||
<td>7.5h</td>
|
||||
</tr>
|
||||
</tbody>
|
||||
</table>
|
||||
</div>
|
||||
|
||||
**Video-MME**
|
||||
|
||||
<div align="left">
|
||||
<table style="margin: 0px auto;">
|
||||
<thead>
|
||||
<tr>
|
||||
<th align="left">Model</th>
|
||||
<th>Size</th>
|
||||
<th>Avg Score ↑</th>
|
||||
<th>Total Inference Time ↓</th>
|
||||
<th>GPU Mem ↓</th>
|
||||
</tr>
|
||||
</thead>
|
||||
<tbody align="center">
|
||||
<tr>
|
||||
<td nowrap="nowrap" align="left">Qwen2.5-VL-7B-Instruct</td>
|
||||
<td>8.3B</td>
|
||||
<td>71.6</td>
|
||||
<td>3h</td>
|
||||
<td>60G</td>
|
||||
</tr>
|
||||
<tr>
|
||||
<td nowrap="nowrap" align="left">GLM-4.1V-9B-Thinking</td>
|
||||
<td>10.3B</td>
|
||||
<td>73.6</td>
|
||||
<td>2.63h</td>
|
||||
<td>32G</td>
|
||||
</tr>
|
||||
<tr>
|
||||
<td nowrap="nowrap" align="left">MiniCPM-V-4_5</td>
|
||||
<td>8.7B</td>
|
||||
<td>73.5</td>
|
||||
<td>0.26h</td>
|
||||
<td>28G</td>
|
||||
</tr>
|
||||
</tbody>
|
||||
</table>
|
||||
</div>
|
||||
|
||||
Both Video-MME and OpenCompass were evaluated using 8×A100 GPUs for inference. The reported inference time of Video-MME excludes the cost of video frame extraction.
|
||||
|
||||
|
||||
### Examples <!-- omit in toc -->
|
||||
|
||||
<div align="center">
|
||||
|
||||
Reference in New Issue
Block a user