Upload folder using huggingface_hub
Browse files
README.md
CHANGED
|
@@ -37,28 +37,28 @@ InternVL 2.0 is a multimodal large language model series, featuring models of va
|
|
| 37 |
|
| 38 |
### Image Benchmarks
|
| 39 |
|
| 40 |
-
|
|
| 41 |
-
|
|
| 42 |
-
|
|
| 43 |
-
|
|
| 44 |
-
|
|
| 45 |
-
|
|
| 46 |
-
|
|
| 47 |
-
|
|
| 48 |
-
|
|
| 49 |
-
|
|
| 50 |
-
|
|
| 51 |
-
|
|
| 52 |
-
|
|
| 53 |
-
|
|
| 54 |
-
|
|
| 55 |
-
|
|
| 56 |
-
|
|
| 57 |
-
|
|
| 58 |
-
|
|
| 59 |
-
|
|
| 60 |
-
|
|
| 61 |
-
|
|
| 62 |
|
| 63 |
- We simultaneously use InternVL and VLMEvalKit repositories for model evaluation. Specifically, the results reported for DocVQA, ChartQA, InfoVQA, TextVQA, MME, AI2D, MMBench, CCBench, MMVet, and SEED-Image were tested using the InternVL repository. OCRBench, RealWorldQA, HallBench, and MathVista were evaluated using the VLMEvalKit.
|
| 64 |
|
|
@@ -66,8 +66,6 @@ InternVL 2.0 is a multimodal large language model series, featuring models of va
|
|
| 66 |
|
| 67 |
- Please note that evaluating the same model using different testing toolkits like InternVL and VLMEvalKit can result in slight differences, which is normal. Updates to code versions and variations in environment and hardware can also cause minor discrepancies in results.
|
| 68 |
|
| 69 |
-
- It is important to mention that the MMVet scores we report are evaluated using GPT-4-0613 as the judge model. Different versions of GPT-4 can lead to significant variations in the scores for this dataset. For instance, using GPT-4-Turbo would result in significantly lower scores.
|
| 70 |
-
|
| 71 |
### Video Benchmarks
|
| 72 |
|
| 73 |
| Benchmark | VideoChat2-Phi3 | VideoChat2-HD-Mistral | Mini-InternVL-4B-1-5 | InternVL2-4B |
|
|
@@ -577,37 +575,35 @@ InternVL 2.0 是一个多模态大语言模型系列，包含各种规模的模
|
|
| 577 |
|
| 578 |
### 图像相关评测
|
| 579 |
|
| 580 |
-
|
|
| 581 |
-
|
|
| 582 |
-
|
|
| 583 |
-
|
|
| 584 |
-
|
|
| 585 |
-
|
|
| 586 |
-
|
|
| 587 |
-
|
|
| 588 |
-
|
|
| 589 |
-
|
|
| 590 |
-
|
|
| 591 |
-
|
|
| 592 |
-
|
|
| 593 |
-
|
|
| 594 |
-
|
|
| 595 |
-
|
|
| 596 |
-
|
|
| 597 |
-
|
|
| 598 |
-
|
|
| 599 |
-
|
|
| 600 |
-
|
|
| 601 |
-
|
|
| 602 |
-
|
| 603 |
-
- 我们同时使用 InternVL 和 VLMEvalKit 仓库进行模型评估。具体来说，DocVQA、ChartQA、InfoVQA、TextVQA、MME、AI2D、MMBench、CCBench、MMVet 和 SEED-Image 的结果是使用 InternVL 仓库测试的。
|
| 604 |
|
| 605 |
- 对于MMMU，我们报告了原始分数（左侧：InternVL系列模型使用InternVL代码库评测，其他模型的分数来自其技术报告或网页）和VLMEvalKit分数（右侧：从OpenCompass排行榜收集）。
|
| 606 |
|
| 607 |
- 请注意，使用不同的测试工具包（如 InternVL 和 VLMEvalKit）评估同一模型可能会导致细微差异，这是正常的。代码版本的更新、环境和硬件的变化也可能导致结果的微小差异。
|
| 608 |
|
| 609 |
-
- 需要提到的是，我们报告的 MMVet 分数是使用 GPT-4-0613 作为评判模型评估的。不同版本的 GPT-4 会导致该数据集分数的显著变化。例如，使用 GPT-4-Turbo 会导致分数显著降低。
|
| 610 |
-
|
| 611 |
### 视频相关评测
|
| 612 |
|
| 613 |
| 评测数据集 | VideoChat2-Phi3 | VideoChat2-HD-Mistral | Mini-InternVL-4B-1-5 | InternVL2-4B |
|
|
|
|
| 37 |
|
| 38 |
### Image Benchmarks
|
| 39 |
|
| 40 |
+
| Benchmark | PaliGemma-3B | Phi-3-Vision | Mini-InternVL-4B-1.5 | InternVL2-4B |
|
| 41 |
+
| :--------------------------: | :----------: | :----------: | :------------------: | :----------: |
|
| 42 |
+
| Model Size | 2.9B | 4.2B | 4.2B | 4.2B |
|
| 43 |
+
| | | | | |
|
| 44 |
+
| DocVQA<sub>test</sub> | - | - | 87.7 | 89.2 |
|
| 45 |
+
| ChartQA<sub>test</sub> | - | 81.4 | 81.0 | 81.5 |
|
| 46 |
+
| InfoVQA<sub>test</sub> | - | - | 64.6 | 67.0 |
|
| 47 |
+
| TextVQA<sub>val</sub> | 68.1 | 70.9 | 72.5 | 74.4 |
|
| 48 |
+
| OCRBench | 614 | 639 | 638 | 788 |
|
| 49 |
+
| MME<sub>sum</sub> | 1686.1 | 1508.0 | 2053.6 | 2064.1 |
|
| 50 |
+
| RealWorldQA | 55.2 | 58.8 | 60.1 | 60.7 |
|
| 51 |
+
| AI2D<sub>test</sub> | 68.3 | 76.7 | 76.9 | 78.9 |
|
| 52 |
+
| MMMU<sub>val</sub> | 34.9 | 40.4 / 46.1 | 43.3 / 45.1 | 47.0 / 48.3 |
|
| 53 |
+
| MMBench-EN<sub>test</sub> | 71.0 | 73.6 | 76.2 | 78.6 |
|
| 54 |
+
| MMBench-CN<sub>test</sub> | 63.6 | - | 70.3 | 73.9 |
|
| 55 |
+
| CCBench<sub>dev</sub> | 29.6 | 24.1 | 58.8 | 66.5 |
|
| 56 |
+
| MMVet<sub>GPT-4-0613</sub> | - | - | 46.7 | 55.7 |
|
| 57 |
+
| MMVet<sub>GPT-4-Turbo</sub> | 33.1 | 44.1 | 43.6 | 51.0 |
|
| 58 |
+
| SEED-Image | 69.6 | 70.9 | 72.5 | 73.7 |
|
| 59 |
+
| HallBench<sub>avg</sub> | 32.2 | 39.0 | 42.8 | 41.9 |
|
| 60 |
+
| MathVista<sub>testmini</sub> | 28.7 | 44.5 | 53.7 | 58.6 |
|
| 61 |
+
| OpenCompass<sub>avg</sub> | 46.6 | 53.6 | 56.2 | 60.6 |
|
| 62 |
|
| 63 |
- We simultaneously use InternVL and VLMEvalKit repositories for model evaluation. Specifically, the results reported for DocVQA, ChartQA, InfoVQA, TextVQA, MME, AI2D, MMBench, CCBench, MMVet, and SEED-Image were tested using the InternVL repository. OCRBench, RealWorldQA, HallBench, and MathVista were evaluated using the VLMEvalKit.
|
| 64 |
|
|
|
|
| 66 |
|
| 67 |
- Please note that evaluating the same model using different testing toolkits like InternVL and VLMEvalKit can result in slight differences, which is normal. Updates to code versions and variations in environment and hardware can also cause minor discrepancies in results.
|
| 68 |
|
|
|
|
|
|
|
| 69 |
### Video Benchmarks
|
| 70 |
|
| 71 |
| Benchmark | VideoChat2-Phi3 | VideoChat2-HD-Mistral | Mini-InternVL-4B-1-5 | InternVL2-4B |
|
|
|
|
| 575 |
|
| 576 |
### 图像相关评测
|
| 577 |
|
| 578 |
+
| 评测数据集 | PaliGemma-3B | Phi-3-Vision | Mini-InternVL-4B-1.5 | InternVL2-4B |
|
| 579 |
+
| :--------------------------: | :----------: | :----------: | :------------------: | :----------: |
|
| 580 |
+
| 模型大小 | 2.9B | 4.2B | 4.2B | 4.2B |
|
| 581 |
+
| | | | | |
|
| 582 |
+
| DocVQA<sub>test</sub> | - | - | 87.7 | 89.2 |
|
| 583 |
+
| ChartQA<sub>test</sub> | - | 81.4 | 81.0 | 81.5 |
|
| 584 |
+
| InfoVQA<sub>test</sub> | - | - | 64.6 | 67.0 |
|
| 585 |
+
| TextVQA<sub>val</sub> | 68.1 | 70.9 | 72.5 | 74.4 |
|
| 586 |
+
| OCRBench | 614 | 639 | 638 | 788 |
|
| 587 |
+
| MME<sub>sum</sub> | 1686.1 | 1508.0 | 2053.6 | 2064.1 |
|
| 588 |
+
| RealWorldQA | 55.2 | 58.8 | 60.1 | 60.7 |
|
| 589 |
+
| AI2D<sub>test</sub> | 68.3 | 76.7 | 76.9 | 78.9 |
|
| 590 |
+
| MMMU<sub>val</sub> | 34.9 | 40.4 / 46.1 | 43.3 / 45.1 | 47.0 / 48.3 |
|
| 591 |
+
| MMBench-EN<sub>test</sub> | 71.0 | 73.6 | 76.2 | 78.6 |
|
| 592 |
+
| MMBench-CN<sub>test</sub> | 63.6 | - | 70.3 | 73.9 |
|
| 593 |
+
| CCBench<sub>dev</sub> | 29.6 | 24.1 | 58.8 | 66.5 |
|
| 594 |
+
| MMVet<sub>GPT-4-0613</sub> | - | - | 46.7 | 55.7 |
|
| 595 |
+
| MMVet<sub>GPT-4-Turbo</sub> | 33.1 | 44.1 | 43.6 | 51.0 |
|
| 596 |
+
| SEED-Image | 69.6 | 70.9 | 72.5 | 73.7 |
|
| 597 |
+
| HallBench<sub>avg</sub> | 32.2 | 39.0 | 42.8 | 41.9 |
|
| 598 |
+
| MathVista<sub>testmini</sub> | 28.7 | 44.5 | 53.7 | 58.6 |
|
| 599 |
+
| OpenCompass<sub>avg</sub> | 46.6 | 53.6 | 56.2 | 60.6 |
|
| 600 |
+
|
| 601 |
+
- 我们同时使用 InternVL 和 VLMEvalKit 仓库进行模型评估。具体来说，DocVQA、ChartQA、InfoVQA、TextVQA、MME、AI2D、MMBench、CCBench、MMVet 和 SEED-Image 的结果是使用 InternVL 仓库测试的。OCRBench、RealWorldQA、HallBench 和 MathVista 是使用 VLMEvalKit 进行评估的。
|
| 602 |
|
| 603 |
- 对于MMMU，我们报告了原始分数（左侧：InternVL系列模型使用InternVL代码库评测，其他模型的分数来自其技术报告或网页）和VLMEvalKit分数（右侧：从OpenCompass排行榜收集）。
|
| 604 |
|
| 605 |
- 请注意，使用不同的测试工具包（如 InternVL 和 VLMEvalKit）评估同一模型可能会导致细微差异，这是正常的。代码版本的更新、环境和硬件的变化也可能导致结果的微小差异。
|
| 606 |
|
|
|
|
|
|
|
| 607 |
### 视频相关评测
|
| 608 |
|
| 609 |
| 评测数据集 | VideoChat2-Phi3 | VideoChat2-HD-Mistral | Mini-InternVL-4B-1-5 | InternVL2-4B |
|