hermanhelf
/

paligemma

@@ -1,199 +1,843 @@
 ---
 library_name: transformers
-tags: []
 ---
-# Model Card for Model ID
-<!-- Provide a quick summary of what the model is/does. -->
-## Model Details
-### Model Description
-<!-- Provide a longer summary of what this model is. -->
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
-- **Developed by:** [More Information Needed]
-- **Funded by [optional]:** [More Information Needed]
-- **Shared by [optional]:** [More Information Needed]
-- **Model type:** [More Information Needed]
-- **Language(s) (NLP):** [More Information Needed]
-- **License:** [More Information Needed]
-- **Finetuned from model [optional]:** [More Information Needed]
-### Model Sources [optional]
-<!-- Provide the basic links for the model. -->
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
-## Uses
-<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
-### Direct Use
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
-[More Information Needed]
-### Downstream Use [optional]
-<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
-[More Information Needed]
-### Out-of-Scope Use
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
-[More Information Needed]
-## Bias, Risks, and Limitations
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
-[More Information Needed]
-### Recommendations
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
-Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
-## How to Get Started with the Model
-Use the code below to get started with the model.
-[More Information Needed]
-## Training Details
-### Training Data
-<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
-[More Information Needed]
-### Training Procedure
-<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
-#### Preprocessing [optional]
-[More Information Needed]
-#### Training Hyperparameters
-- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
-#### Speeds, Sizes, Times [optional]
-<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
-[More Information Needed]
-## Evaluation
-<!-- This section describes the evaluation protocols and provides the results. -->
-### Testing Data, Factors & Metrics
-#### Testing Data
-<!-- This should link to a Dataset Card if possible. -->
-[More Information Needed]
-#### Factors
-<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
-[More Information Needed]
-#### Metrics
-<!-- These are the evaluation metrics being used, ideally with a description of why. -->
-[More Information Needed]
-### Results
-[More Information Needed]
-#### Summary
-## Model Examination [optional]
-<!-- Relevant interpretability work for the model goes here -->
-[More Information Needed]
-## Environmental Impact
-<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
-Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
-- **Hardware Type:** [More Information Needed]
-- **Hours used:** [More Information Needed]
-- **Cloud Provider:** [More Information Needed]
-- **Compute Region:** [More Information Needed]
-- **Carbon Emitted:** [More Information Needed]
-## Technical Specifications [optional]
-### Model Architecture and Objective
-[More Information Needed]
-### Compute Infrastructure
-[More Information Needed]
-#### Hardware
-[More Information Needed]
-#### Software
-[More Information Needed]
-## Citation [optional]
-<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
-**BibTeX:**
-[More Information Needed]
-**APA:**
-[More Information Needed]
-## Glossary [optional]
-<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
-[More Information Needed]
-## More Information [optional]
-[More Information Needed]
-## Model Card Authors [optional]
-[More Information Needed]
-## Model Card Contact
-[More Information Needed]

 ---
+license: gemma
 library_name: transformers
+extra_gated_heading: Access PaliGemma on Hugging Face
+extra_gated_prompt: To access PaliGemma on Hugging Face, you’re required to review
+  and agree to Google’s usage license. To do this, please ensure you’re logged-in
+  to Hugging Face and click below. Requests are processed immediately.
+extra_gated_button_content: Acknowledge license
+pipeline_tag: image-text-to-text
 ---
+# PaliGemma model card
+**Model page:** [PaliGemma](https://ai.google.dev/gemma/docs/paligemma)
+Transformers PaliGemma 3B weights, pre-trained with 224*224 input images and 128 token input/output text sequences. The models are available in float32, bfloat16 and float16 formats for fine-tuning.
+**Resources and technical documentation:**
+*   [Responsible Generative AI Toolkit](https://ai.google.dev/responsible)
+*   [PaliGemma on Kaggle](https://www.kaggle.com/models/google/paligemma)
+*   [PaliGemma on Vertex Model Garden](https://console.cloud.google.com/vertex-ai/publishers/google/model-garden/363)
+**Terms of Use:** [Terms](https://ai.google.dev/gemma/terms)
+**Authors:** Google
+## Model information
+### Model summary
+#### Description
+PaliGemma is a versatile and lightweight vision-language model (VLM) inspired by
+[PaLI-3](https://arxiv.org/abs/2310.09199) and based on open components such as
+the [SigLIP vision model](https://arxiv.org/abs/2303.15343) and the [Gemma
+language model](https://arxiv.org/abs/2403.08295). It takes both image and text
+as input and generates text as output, supporting multiple languages. It is designed for class-leading fine-tune performance on a wide range of vision-language tasks such as image and short video caption, visual question answering, text reading, object detection and object segmentation.
+#### Model architecture
+PaliGemma is the composition of a [Transformer
+decoder](https://arxiv.org/abs/1706.03762) and a [Vision Transformer image
+encoder](https://arxiv.org/abs/2010.11929), with a total of 3 billion
+params. The text decoder is initialized from
+[Gemma-2B](https://www.kaggle.com/models/google/gemma). The image encoder is
+initialized from
+[SigLIP-So400m/14](https://colab.research.google.com/github/google-research/big_vision/blob/main/big_vision/configs/proj/image_text/SigLIP_demo.ipynb).
+PaliGemma is trained following the PaLI-3 recipes.
+#### Inputs and outputs
+*   **Input:** Image and text string, such as a prompt to caption the image, or
+    a question.
+*   **Output:** Generated text in response to the input, such as a caption of
+    the image, an answer to a question, a list of object bounding box
+    coordinates, or segmentation codewords.
+### Model data
+#### Pre-train datasets
+PaliGemma is pre-trained on the following mixture of datasets:
+*   **WebLI:** [WebLI (Web Language Image)](https://arxiv.org/abs/2209.06794) is
+    a web-scale multilingual image-text dataset built from the public web. A
+    wide range of WebLI splits are used to acquire versatile model capabilities,
+    such as visual semantic understanding, object localization,
+    visually-situated text understanding, multilinguality, etc.
+*   **CC3M-35L:** Curated English image-alt_text pairs from webpages ([Sharma et
+    al., 2018](https://aclanthology.org/P18-1238/)). We used the [Google Cloud
+    Translation API](https://cloud.google.com/translate) to translate into 34
+    additional languages.
+*   **VQ²A-CC3M-35L/VQG-CC3M-35L:** A subset of VQ2A-CC3M ([Changpinyo et al.,
+    2022a](https://aclanthology.org/2022.naacl-main.142/)), translated into the
+    same additional 34 languages as CC3M-35L, using the [Google Cloud
+    Translation API](https://cloud.google.com/translate).
+*   **OpenImages:** Detection and object-aware questions and answers
+    ([Piergiovanni et al. 2022](https://arxiv.org/abs/2209.04372)) generated by
+    handcrafted rules on the [OpenImages dataset].
+*   **WIT:** Images and texts collected from Wikipedia ([Srinivasan et al.,
+    2021](https://arxiv.org/abs/2103.01913)).
+[OpenImages dataset]: https://storage.googleapis.com/openimages/web/factsfigures_v7.html
+#### Data responsibility filtering
+The following filters are applied to WebLI, with the goal of training PaliGemma
+on clean data:
+*   **Pornographic image filtering:** This filter removes images deemed to be of
+    pornographic nature.
+*   **Text safety filtering:** We identify and filter out images that are paired
+    with unsafe text. Unsafe text is any text deemed to contain or be about
+    CSAI, pornography, vulgarities, or otherwise offensive.
+*   **Text toxicity filtering:** We further use the [Perspective
+    API](https://perspectiveapi.com/) to identify and filter out images that are
+    paired with text deemed insulting, obscene, hateful or otherwise toxic.
+*   **Text personal information filtering:** We filtered certain personal information and other sensitive data using [Cloud Data Loss Prevention (DLP)
+    API](https://cloud.google.com/security/products/dlp) to protect the privacy
+    of individuals. Identifiers such as social security numbers and [other sensitive information types] were removed.
+*   **Additional methods:** Filtering based on content quality and safety in
+    line with our policies and practices.
+[other sensitive information types]: https://cloud.google.com/sensitive-data-protection/docs/high-sensitivity-infotypes-reference?_gl=1*jg604m*_ga*ODk5MzA3ODQyLjE3MTAzMzQ3NTk.*_ga_WH2QY8WWF5*MTcxMDUxNTkxMS4yLjEuMTcxMDUxNjA2NC4wLjAuMA..&_ga=2.172110058.-899307842.1710334759
+## How to Use
+PaliGemma is a single-turn vision language model not meant for conversational use,
+and it works best when fine-tuning to a specific use case.
+You can configure which task the model will solve by conditioning it with task prefixes,
+such as “detect” or “segment”. The pretrained models were trained in this fashion to imbue
+them with a rich set of capabilities (question answering, captioning, segmentation, etc.).
+However, they are not designed to be used directly, but to be transferred (by fine-tuning)
+to specific tasks using a similar prompt structure. For interactive testing, you can use
+the "mix" family of models, which have been fine-tuned on a mixture of tasks.
+Please, refer to the [usage and limitations section](#usage-and-limitations) for intended
+use cases, or visit the [blog post](https://huggingface.co/blog/paligemma-google-vlm) for
+additional details and examples.
+## Use in Transformers
+The following snippets use model `google/paligemma-3b-mix-224` for reference purposes.
+The model in this repo you are now browsing may have been trained for other tasks, please
+make sure you use appropriate inputs for the task at hand.
+### Running the default precision (`float32`) on CPU
+```python
+from transformers import AutoProcessor, PaliGemmaForConditionalGeneration
+from PIL import Image
+import requests
+import torch
+model_id = "google/paligemma-3b-mix-224"
+url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/car.jpg?download=true"
+image = Image.open(requests.get(url, stream=True).raw)
+model = PaliGemmaForConditionalGeneration.from_pretrained(model_id).eval()
+processor = AutoProcessor.from_pretrained(model_id)
+# Instruct the model to create a caption in Spanish
+prompt = "caption es"
+model_inputs = processor(text=prompt, images=image, return_tensors="pt")
+input_len = model_inputs["input_ids"].shape[-1]
+with torch.inference_mode():
+    generation = model.generate(**model_inputs, max_new_tokens=100, do_sample=False)
+    generation = generation[0][input_len:]
+    decoded = processor.decode(generation, skip_special_tokens=True)
+    print(decoded)
+```
+Output: `Un auto azul estacionado frente a un edificio.`
+### Running other precisions on CUDA
+For convenience, the repos contain revisions of the weights already converted to `bfloat16` and `float16`,
+so you can use them to reduce the download size and avoid casting on your local computer.
+This is how you'd run `bfloat16` on an nvidia CUDA card.
+```python
+from transformers import AutoProcessor, PaliGemmaForConditionalGeneration
+from PIL import Image
+import requests
+import torch
+model_id = "google/paligemma-3b-mix-224"
+device = "cuda:0"
+dtype = torch.bfloat16
+url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/car.jpg?download=true"
+image = Image.open(requests.get(url, stream=True).raw)
+model = PaliGemmaForConditionalGeneration.from_pretrained(
+    model_id,
+    torch_dtype=dtype,
+    device_map=device,
+    revision="bfloat16",
+).eval()
+processor = AutoProcessor.from_pretrained(model_id)
+# Instruct the model to create a caption in Spanish
+prompt = "caption es"
+model_inputs = processor(text=prompt, images=image, return_tensors="pt").to(model.device)
+input_len = model_inputs["input_ids"].shape[-1]
+with torch.inference_mode():
+    generation = model.generate(**model_inputs, max_new_tokens=100, do_sample=False)
+    generation = generation[0][input_len:]
+    decoded = processor.decode(generation, skip_special_tokens=True)
+    print(decoded)
+```
+### Loading in 4-bit / 8-bit
+You need to install `bitsandbytes` to automatically run inference using 8-bit or 4-bit precision:
+```
+pip install bitsandbytes accelerate
+```
+```
+from transformers import AutoProcessor, PaliGemmaForConditionalGeneration
+from PIL import Image
+import requests
+import torch
+model_id = "google/paligemma-3b-mix-224"
+device = "cuda:0"
+dtype = torch.bfloat16
+url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/car.jpg?download=true"
+image = Image.open(requests.get(url, stream=True).raw)
+quantization_config = BitsAndBytesConfig(load_in_8bit=True)
+model = PaliGemmaForConditionalGeneration.from_pretrained(
+    model_id, quantization_config=quantization_config
+).eval()
+processor = AutoProcessor.from_pretrained(model_id)
+# Instruct the model to create a caption in Spanish
+prompt = "caption es"
+model_inputs = processor(text=prompt, images=image, return_tensors="pt").to(model.device)
+input_len = model_inputs["input_ids"].shape[-1]
+with torch.inference_mode():
+    generation = model.generate(**model_inputs, max_new_tokens=100, do_sample=False)
+    generation = generation[0][input_len:]
+    decoded = processor.decode(generation, skip_special_tokens=True)
+    print(decoded)
+```
+## Implementation information
+### Hardware
+PaliGemma was trained using the latest generation of Tensor Processing Unit
+(TPU) hardware (TPUv5e).
+### Software
+Training was done using [JAX](https://github.com/google/jax),
+[Flax](https://github.com/google/flax),
+[TFDS](https://github.com/tensorflow/datasets) and
+[`big_vision`](https://github.com/google-research/big_vision).
+JAX allows researchers to take advantage of the latest generation of hardware,
+including TPUs, for faster and more efficient training of large models.
+TFDS is used to access datasets and Flax is used for model architecture. The
+PaliGemma fine-tune code and inference code are released in the `big_vision`
+GitHub repository.
+## Evaluation information
+### Benchmark results
+In order to verify the transferability of PaliGemma to a wide variety of
+academic tasks, we fine-tune the pretrained models on each task. Additionally we
+train the mix model with a mixture of the transfer tasks. We report results on
+different resolutions to provide an impression of which tasks benefit from
+increased resolution. Importantly, none of these tasks or datasets are part of
+the pretraining data mixture, and their images are explicitly removed from the
+web-scale pre-training data.
+#### Single task (fine-tune on single task)
+<table>
+  <tbody><tr>
+    <th>Benchmark<br>(train split)</th>
+    <th>Metric<br>(split)</th>
+    <th>pt-224</th>
+    <th>pt-448</th>
+    <th>pt-896</th>
+  </tr>
+  <tr>
+    <th>Captioning</th>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://cocodataset.org/#home">COCO captions</a><br>(train+restval)
+    </td>
+    <td>CIDEr (val)</td>
+    <td>141.92</td>
+    <td>144.60</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://nocaps.org/">NoCaps</a><br>(Eval of COCO<br>captions transfer)
+    </td>
+    <td>CIDEr (val)</td>
+    <td>121.72</td>
+    <td>123.58</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/pdf/2205.12522">COCO-35L</a><br>(train)
+    </td>
+    <td>CIDEr dev<br>(en/avg-34/avg)</td>
+    <td>
+      139.2<br>
+      115.8<br>
+      116.4
+    </td>
+    <td>
+      141.2<br>
+      118.0<br>
+      118.6
+    </td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/pdf/2205.12522">XM3600</a><br>(Eval of COCO-35L transfer)
+    </td>
+    <td>CIDEr dev<br>(en/avg-34/avg)</td>
+    <td>
+      78.1<br>
+      41.3<br>
+      42.4
+    </td>
+    <td>
+      80.0<br>
+      41.9<br>
+      42.9
+    </td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://textvqa.org/textcaps/">TextCaps</a><br>(train)
+    </td>
+    <td>CIDEr (val)</td>
+    <td>127.48</td>
+    <td>153.94</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/abs/2110.11624">SciCap</a><br>(first sentence, no subfigure)<br>(train+val)
+    </td>
+    <td>CIDEr/BLEU-4<br>(test)</td>
+    <td>
+      162.25<br>
+      0.192<br>
+    </td>
+    <td>
+      181.49<br>
+      0.211<br>
+    </td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/abs/2108.03353">Screen2words</a><br>(train+dev)
+    </td>
+    <td>CIDEr (test)</td>
+    <td>117.57</td>
+    <td>119.59</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/abs/2010.04295">Widget Captioning</a><br>(train+dev)
+    </td>
+    <td>CIDEr (test)</td>
+    <td>136.07</td>
+    <td>148.36</td>
+  </tr>
+  <tr>
+    <th>Question answering</th>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://visualqa.org/index.html">VQAv2</a><br>(train+validation)
+    </td>
+    <td>Accuracy<br>(Test server - std)</td>
+    <td>83.19</td>
+    <td>85.64</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/abs/2401.06209">MMVP</a><br>(Eval of VQAv2 transfer)
+    </td>
+    <td>Paired Accuracy</td>
+    <td>47.33</td>
+    <td>45.33</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/abs/2305.10355">POPE</a><br>(Eval of VQAv2 transfer)
+    </td>
+    <td>Accuracy<br>(random/popular/<br>adversarial)</td>
+    <td>
+      87.80<br>
+      85.87<br>
+      84.27
+    </td>
+    <td>
+      88.23<br>
+      86.77<br>
+      85.90
+    </td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://okvqa.allenai.org/">OKVQA</a><br>(train)
+    </td>
+    <td>Accuracy (val)</td>
+    <td>63.54</td>
+    <td>63.15</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://allenai.org/project/a-okvqa/home">A-OKVQA</a> (MC)<br>(train+val)
+    </td>
+    <td>Accuracy<br>(Test server)</td>
+    <td>76.37</td>
+    <td>76.90</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://allenai.org/project/a-okvqa/home">A-OKVQA</a> (DA)<br>(train+val)
+    </td>
+    <td>Accuracy<br>(Test server)</td>
+    <td>61.85</td>
+    <td>63.22</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://cs.stanford.edu/people/dorarad/gqa/about.html">GQA</a><br>(train_balanced+<br>val_balanced)
+    </td>
+    <td>Accuracy<br>(testdev balanced)</td>
+    <td>65.61</td>
+    <td>67.03</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://aclanthology.org/2022.findings-acl.196/">xGQA</a><br>(Eval of GQA transfer)
+    </td>
+    <td>Mean Accuracy<br>(bn, de, en, id,<br>ko, pt, ru, zh)</td>
+    <td>58.37</td>
+    <td>59.07</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://lil.nlp.cornell.edu/nlvr/">NLVR2</a><br>(train+dev)
+    </td>
+    <td>Accuracy (test)</td>
+    <td>90.02</td>
+    <td>88.93</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://marvl-challenge.github.io/">MaRVL</a><br>(Eval of NLVR2 transfer)
+    </td>
+    <td>Mean Accuracy<br>(test)<br>(id, sw, ta, tr, zh)</td>
+    <td>80.57</td>
+    <td>76.78</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://allenai.org/data/diagrams">AI2D</a><br>(train)
+    </td>
+    <td>Accuracy (test)</td>
+    <td>72.12</td>
+    <td>73.28</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://scienceqa.github.io/">ScienceQA</a><br>(Img subset, no CoT)<br>(train+val)
+    </td>
+    <td>Accuracy (test)</td>
+    <td>95.39</td>
+    <td>95.93</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://zenodo.org/records/6344334">RSVQA-LR</a> (Non numeric)<br>(train+val)
+    </td>
+    <td>Mean Accuracy<br>(test)</td>
+    <td>92.65</td>
+    <td>93.11</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://zenodo.org/records/6344367">RSVQA-HR</a> (Non numeric)<br>(train+val)
+    </td>
+    <td>Mean Accuracy<br>(test/test2)</td>
+    <td>
+      92.61<br>
+      90.58
+    </td>
+    <td>
+      92.79<br>
+      90.54
+    </td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/abs/2203.10244">ChartQA</a><br>(human+aug)x(train+val)
+    </td>
+    <td>Mean Relaxed<br>Accuracy<br>(test_human,<br>test_aug)</td>
+    <td>57.08</td>
+    <td>71.36</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://vizwiz.org/tasks-and-datasets/vqa/">VizWiz VQA</a><br>(train+val)
+    </td>
+    <td>Accuracy<br>(Test server - std)</td>
+    <td>
+      73.7
+    </td>
+    <td>
+      75.52
+    </td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/abs/1810.12440">TallyQA</a><br>(train)
+    </td>
+    <td>Accuracy<br>(test_simple/<br>test_complex)</td>
+    <td>
+      81.72<br>
+      69.56
+    </td>
+    <td>
+      84.86<br>
+      72.27
+    </td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://ocr-vqa.github.io/">OCR-VQA</a><br>(train+val)
+    </td>
+    <td>Accuracy (test)</td>
+    <td>72.32</td>
+    <td>74.61</td>
+    <td>74.93</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://textvqa.org/">TextVQA</a><br>(train+val)
+    </td>
+    <td>Accuracy<br>(Test server - std)</td>
+    <td>55.47</td>
+    <td>73.15</td>
+    <td>76.48</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://www.docvqa.org/">DocVQA</a><br>(train+val)
+    </td>
+    <td>ANLS (Test server)</td>
+    <td>43.74</td>
+    <td>78.02</td>
+    <td>84.77</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://openaccess.thecvf.com/content/WACV2022/papers/Mathew_InfographicVQA_WACV_2022_paper.pdf">Infographic VQA</a><br>(train+val)
+    </td>
+    <td>ANLS (Test server)</td>
+    <td>28.46</td>
+    <td>40.47</td>
+    <td>47.75</td>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/abs/1905.13648">SceneText VQA</a><br>(train+val)
+    </td>
+    <td>ANLS (Test server)</td>
+    <td>63.29</td>
+    <td>81.82</td>
+    <td>84.40</td>
+  </tr>
+  <tr>
+    <th>Segmentation</th>
+  </tr>
+  <tr>
+    <td>
+      <a href="https://arxiv.org/abs/1608.00272">RefCOCO</a><br>(combined refcoco, refcoco+,<br>refcocog excluding val<br>and test images)
+    </td>
+    <td>MIoU<br>(validation)<br>refcoco/refcoco+/<br>refcocog</td>
+    <td>
+      73.40<br>
+      68.32<br>
+      67.65
+    </td>
+    <td>
+      75.57<br>
+      69.76<br>
+      70.17
+    </td>
+    <td>
+      76.94<br>
+      72.18<br>
+      72.22
+    </td>
+  </tr>
+  <tr>
+    <th>Video tasks (Caption/QA)</th>
+  </tr>
+  <tr>
+    <td>MSR-VTT (Captioning)</td>
+    <td>CIDEr (test)</td>
+    <td>70.54</td>
+  </tr>
+  <tr>
+    <td>MSR-VTT (QA)</td>
+    <td>Accuracy (test)</td>
+    <td>50.09</td>
+  </tr>
+  <tr>
+    <td>ActivityNet (Captioning)</td>
+    <td>CIDEr (test)</td>
+    <td>34.62</td>
+  </tr>
+  <tr>
+    <td>ActivityNet (QA)</td>
+    <td>Accuracy (test)</td>
+    <td>50.78</td>
+  </tr>
+  <tr>
+    <td>VATEX (Captioning)</td>
+    <td>CIDEr (test)</td>
+    <td>79.73</td>
+  </tr>
+  <tr>
+    <td>MSVD (QA)</td>
+    <td>Accuracy (test)</td>
+    <td>60.22</td>
+  </tr>
+</tbody></table>
+#### Mix model (fine-tune on mixture of transfer tasks)
+<table>
+  <tbody><tr>
+    <th>Benchmark</th>
+    <th>Metric (split)</th>
+    <th>mix-224</th>
+    <th>mix-448</th>
+  </tr>
+  <tr>
+    <td><a href="https://arxiv.org/abs/2401.06209">MMVP</a></td>
+    <td>Paired Accuracy</td>
+    <td>46.00</td>
+    <td>45.33</td>
+  </tr>
+  <tr>
+    <td><a href="https://arxiv.org/abs/2305.10355">POPE</a></td>
+    <td>Accuracy<br>(random/popular/adversarial)</td>
+    <td>
+      88.00<br>
+      86.63<br>
+      85.67
+    </td>
+    <td>
+      89.37<br>
+      88.40<br>
+      87.47
+    </td>
+  </tr>
+</tbody></table>
+## Ethics and safety
+### Evaluation approach
+Our evaluation methods include structured evaluations and internal red-teaming
+testing of relevant content policies. Red-teaming was conducted by a number of
+different teams, each with different goals and human evaluation metrics. These
+models were evaluated against a number of different categories relevant to
+ethics and safety, including:
+*   Human evaluation on prompts covering child safety, content safety and
+    representational harms. See the [Gemma model
+    card](https://ai.google.dev/gemma/docs/model_card#evaluation_approach) for
+    more details on evaluation approach, but with image captioning and visual
+    question answering setups.
+*   Image-to-Text benchmark evaluation: Benchmark against relevant academic
+    datasets such as FairFace Dataset ([Karkkainen et al.,
+    2021](https://arxiv.org/abs/1908.04913)).
+### Evaluation results
+*   The human evaluation results of ethics and safety evaluations are within
+    acceptable thresholds for meeting [internal
+    policies](https://storage.googleapis.com/gweb-uniblog-publish-prod/documents/2023_Google_AI_Principles_Progress_Update.pdf#page=11)
+    for categories such as child safety, content safety and representational
+    harms.
+*   On top of robust internal evaluations, we also use the Perspective API
+    (threshold of 0.8) to measure toxicity, profanity, and other potential
+    issues in the generated captions for images sourced from the FairFace
+    dataset. We report the maximum and median values observed across subgroups
+    for each of the perceived gender, ethnicity, and age attributes.
+<table>
+  <tbody><tr>
+    </tr></tbody><tbody><tr><th>Metric</th>
+    <th>Perceived<br>gender</th>
+    <th></th>
+    <th>Ethnicity</th>
+    <th></th>
+    <th>Age group</th>
+    <th></th>
+  </tr>
+  <tr>
+    <th></th>
+    <th>Maximum</th>
+    <th>Median</th>
+    <th>Maximum</th>
+    <th>Median</th>
+    <th>Maximum</th>
+    <th>Median</th>
+  </tr>
+  <tr>
+    <td>Toxicity</td>
+    <td>0.04%</td>
+    <td>0.03%</td>
+    <td>0.08%</td>
+    <td>0.00%</td>
+    <td>0.09%</td>
+    <td>0.00%</td>
+  </tr>
+  <tr>
+    <td>Identity Attack</td>
+    <td>0.00%</td>
+    <td>0.00%</td>
+    <td>0.00%</td>
+    <td>0.00%</td>
+    <td>0.00%</td>
+    <td>0.00%</td>
+  </tr>
+  <tr>
+    <td>Insult</td>
+    <td>0.06%</td>
+    <td>0.04%</td>
+    <td>0.09%</td>
+    <td>0.07%</td>
+    <td>0.16%</td>
+    <td>0.00%</td>
+  </tr>
+  <tr>
+    <td>Threat</td>
+    <td>0.06%</td>
+    <td>0.05%</td>
+    <td>0.14%</td>
+    <td>0.05%</td>
+    <td>0.17%</td>
+    <td>0.00%</td>
+  </tr>
+  <tr>
+    <td>Profanity</td>
+    <td>0.00%</td>
+    <td>0.00%</td>
+    <td>0.00%</td>
+    <td>0.00%</td>
+    <td>0.00%</td>
+    <td>0.00%</td>
+  </tr>
+</tbody></table>
+## Usage and limitations
+### Intended usage
+Open Vision Language Models (VLMs) have a wide range of applications across
+various industries and domains. The following list of potential uses is not
+comprehensive. The purpose of this list is to provide contextual information
+about the possible use-cases that the model creators considered as part of model
+training and development.
+Fine-tune on specific vision-language task:
+*   The pre-trained models can be fine-tuned on a wide range of vision-language
+    tasks such as: image captioning, short video caption, visual question
+    answering, text reading, object detection and object segmentation.
+*   The pre-trained models can be fine-tuned for specific domains such as remote
+    sensing question answering, visual questions from people who are blind,
+    science question answering, describe UI element functionalities.
+*   The pre-trained models can be fine-tuned for tasks with non-textual outputs
+    such as bounding boxes or segmentation masks.
+Vision-language research:
+*   The pre-trained models and fine-tuned models can serve as a foundation for researchers to experiment with VLM
+    techniques, develop algorithms, and contribute to the advancement of the
+    field.
+### Ethical considerations and risks
+The development of vision-language models (VLMs) raises several ethical concerns. In creating an open model, we have carefully considered the following:
+*   Bias and Fairness
+    *   VLMs trained on large-scale, real-world image-text data can reflect socio-cultural biases embedded in the training material. These models underwent careful scrutiny, input data pre-processing described and posterior evaluations reported in this card.
+*   Misinformation and Misuse
+    *   VLMs can be misused to generate text that is false, misleading, or harmful.
+    *   Guidelines are provided for responsible use with the model, see the [Responsible Generative AI Toolkit](https://ai.google.dev/responsible).
+*   Transparency and Accountability
+    *   This model card summarizes details on the models' architecture, capabilities, limitations, and evaluation processes.
+    *   A responsibly developed open model offers the opportunity to share innovation by making VLM technology accessible to developers and researchers across the AI ecosystem.
+Risks identified and mitigations:
+*   **Perpetuation of biases:** It's encouraged to perform continuous monitoring
+    (using evaluation metrics, human review) and the exploration of de-biasing
+    techniques during model training, fine-tuning, and other use cases.
+*   **Generation of harmful content:** Mechanisms and guidelines for content
+    safety are essential. Developers are encouraged to exercise caution and
+    implement appropriate content safety safeguards based on their specific
+    product policies and application use cases.
+*   **Misuse for malicious purposes:** Technical limitations and developer and
+    end-user education can help mitigate against malicious applications of LLMs.
+    Educational resources and reporting mechanisms for users to flag misuse are
+    provided. Prohibited uses of Gemma models are outlined in the [Gemma
+    Prohibited Use Policy](https://ai.google.dev/gemma/prohibited_use_policy).
+*   **Privacy violations:** Models were trained on data filtered to remove certain personal information and sensitive data. Developers are encouraged to adhere to privacy regulations with privacy-preserving techniques.
+### Limitations
+*   Most limitations inherited from the underlying Gemma model still apply:
+    *   VLMs are better at tasks that can be framed with clear prompts and
+        instructions. Open-ended or highly complex tasks might be challenging.
+    *   Natural language is inherently complex. VLMs might struggle to grasp
+        subtle nuances, sarcasm, or figurative language.
+    *   VLMs generate responses based on information they learned from their
+        training datasets, but they are not knowledge bases. They may generate
+        incorrect or outdated factual statements.
+    *   VLMs rely on statistical patterns in language and images. They might
+        lack the ability to apply common sense reasoning in certain situations.
+*   PaliGemma was designed first and foremost to serve as a general pre-trained
+    model for transfer to specialized tasks. Hence, its "out of the box" or
+    "zero-shot" performance might lag behind models designed specifically for
+    that.
+*   PaliGemma is not a multi-turn chatbot. It is designed for a single round of
+    image and text input.