Merge pull request facebookresearch#124 from vkehfdl1/dockerfile-api

Dockerfile for running nougat FastAPI server

docker/Dockerfile

@@ -0,0 +1,24 @@
+FROM nvidia/cuda:11.8.0-cudnn8-devel-ubuntu22.04
+# replace CUDA version to your CUDA version.
+# You can check your CUDA version with below.
+# nvcc -V
+
+RUN apt-get update
+RUN apt-get install -y python3
+RUN apt-get -y install python3-pip git
+RUN pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
+# replace CUDA version to your CUDA version.
+
+RUN mkdir workspace
+WORKDIR /workspace
+
+RUN pip3 install fastapi uvicorn[standard] fsspec[http]==2023.1.0
+RUN git clone https://github.com/facebookresearch/nougat.git
+WORKDIR /workspace/nougat
+
+RUN python3 setup.py install
+
+EXPOSE 8503
+
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "8503"]
+# Run this using 'docker run -it -d -p <YOUR PORT>:8503 --gpus all <IMAGE NAME>

docker/README.md

@@ -0,0 +1,60 @@
+## Prerequisites
+Ensure you have Docker installed on your machine. 
+And you must also have NVIDIA CUDA and CuDNN installed in your machine. 
+
+Then, you must check your machine's CUDA version.
+```sh
+nvcc -V
+```
+
+You must change base image name and pytorch version compatible with your **CUDA version**. 
+
+## Building the Docker Image
+Clone this repository and navigate into the current directory(nougat/docker). You can build the Docker image by running:
+```sh
+docker build -t <image-name> .
+```
+Replace <image-name> with a name of your choice. This will be used to refer to the image later.
+Please be patient as this operation can take a while. It needs to pull the CUDA-capable image from NVIDIA’s Docker repository and install several libraries.
+Image size will be about 17GB.
+
+
+## Running the Docker Container
+You can run your Docker container with the following command:
+```sh
+docker run -it -d -p <your-port>:8503 --gpus all <image-name>
+```
+Replace <your-port> with the port number you wish to expose on your host machine to access the nougat API server.
+This can be any valid port number. Replace <image-name> with the name you chose earlier during the build step.
+
+
+## Testing the API Server
+Once the Docker container is running, you can access the nougat API server.
+You can easily check connection by running:
+```sh
+curl -X 'GET' \
+  'http://127.0.0.1:<your-port>/'
+```
+It can be take a while for loading API server, because the server have to download nougat model at startup.
+
+If connection is successful, you can get response looks like this.
+```
+{"status-code":200,"data":{}}
+```
+
+## Using the API Server
+To get a prediction of a PDF file by making a POST request to `http://127.0.0.1:<your-port>/predict/`. It also accepts parameters `start` and `stop` to limit the computation to select page numbers (boundaries are included).
+
+The response is a string with the markdown text of the document.
+
+```sh
+curl -X 'POST' \
+  'http://127.0.0.1:<your-port>/predict/' \
+  -H 'accept: application/json' \
+  -H 'Content-Type: multipart/form-data' \
+  -F 'file=@<PDFFILE.pdf>;type=application/pdf'
+```
+To use the limit the conversion to pages 1 to 5, use the start/stop parameters in the request URL: 
+`http://127.0.0.1:<your-port>/predict/?start=1&stop=5`
+
+