add vcr results
Browse files
README.md
CHANGED
@@ -52,13 +52,15 @@ We have three models with 2, 7 and 72 billion parameters. This repo contains the
|
|
52 |
|
53 |
| Benchmark | InternVL2-2B | MiniCPM-V 2.0 | **Qwen2-VL-2B** |
|
54 |
| :--- | :---: | :---: | :---: |
|
|
|
55 |
| DocVQA<sub>test</sub> | 86.9 | - | **90.1** |
|
56 |
| InfoVQA<sub>test</sub> | 58.9 | - | **65.5** |
|
57 |
| ChartQA<sub>test</sub> | **76.2** | - | 73.5 |
|
58 |
| TextVQA<sub>val</sub> | 73.4 | - | **79.7** |
|
59 |
| OCRBench | 781 | 605 | **794** |
|
60 |
| MTVQA | - | - | **20.0** |
|
61 |
-
|
|
|
|
62 |
| RealWorldQA | 57.3 | 55.8 | **62.9** |
|
63 |
| MME<sub>sum</sub> | **1876.8** | 1808.6 | 1872.0 |
|
64 |
| MMBench-EN<sub>test</sub> | 73.2 | 69.1 | **74.9** |
|
|
|
52 |
|
53 |
| Benchmark | InternVL2-2B | MiniCPM-V 2.0 | **Qwen2-VL-2B** |
|
54 |
| :--- | :---: | :---: | :---: |
|
55 |
+
| MMMU<sub>val</sub> | 36.3 | 38.2 | **41.1** |
|
56 |
| DocVQA<sub>test</sub> | 86.9 | - | **90.1** |
|
57 |
| InfoVQA<sub>test</sub> | 58.9 | - | **65.5** |
|
58 |
| ChartQA<sub>test</sub> | **76.2** | - | 73.5 |
|
59 |
| TextVQA<sub>val</sub> | 73.4 | - | **79.7** |
|
60 |
| OCRBench | 781 | 605 | **794** |
|
61 |
| MTVQA | - | - | **20.0** |
|
62 |
+
| VCR<sub>en easy</sub> | - | - | **81.45**
|
63 |
+
| VCR<sub>zh easy</sub> | - | - | **46.16**
|
64 |
| RealWorldQA | 57.3 | 55.8 | **62.9** |
|
65 |
| MME<sub>sum</sub> | **1876.8** | 1808.6 | 1872.0 |
|
66 |
| MMBench-EN<sub>test</sub> | 73.2 | 69.1 | **74.9** |
|