Merge pull request #21 from debanjum/saba/dockerize

Add Docker support to semantic-search
This commit is contained in:
Debanjum 2022-01-28 20:27:40 -08:00 committed by GitHub
commit d943d2be80
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
6 changed files with 169 additions and 31 deletions

29
Dockerfile Normal file
View file

@ -0,0 +1,29 @@
# syntax=docker/dockerfile:1
FROM continuumio/miniconda3:latest
# Install system dependencies.
RUN apt-get update -y && \
apt-get -y install libimage-exiftool-perl
# Add the local code to the /app directory and set it to be the working directory.
# Since we mount the /app directory as a volume in docker-compose.yml, this
# allows us to automatically update the code in the Docker image when it's changed.
ADD . /app
WORKDIR /app
# Get the arguments from the docker-compose environment.
ARG PORT
EXPOSE ${PORT}
# Create the conda environment.
RUN conda env create -f environment.yml
# Use the conda environment we created to run the application.
# To enable the conda env, we cannot simply RUN `conda activate semantic-search`,
# since each RUN command in a Dockerfile is a separate bash shell.
# The environment would not carry forward.
# Instead, we'll use `conda run` to run the application.
# There are more arguments required for the script to run,
# but these should be passed in through the docker-compose.yml file.
ENTRYPOINT ["conda", "run", "--no-capture-output", "--name", "semantic-search", \
"python3", "-m", "src.main"]

View file

@ -5,32 +5,52 @@
All data is processed locally. User can interface with semantic-search app via [[./src/interface/emacs/semantic-search.el][Emacs]], API or Commandline
** Dependencies
- Python3
- [[https://docs.conda.io/en/latest/miniconda.html#latest-miniconda-installer-links][Miniconda]]
** Setup
** Install
#+begin_src shell
git clone https://github.com/debanjum/semantic-search && cd semantic-search
conda env create -f environment.yml
conda activate semantic-search
#+end_src
*** Setup using Docker
*** Install Environmental Dependencies
#+begin_src shell
sudo apt-get -y install libimage-exiftool-perl
#+end_src
**** 1. Clone Repository
#+begin_src shell
git clone https://github.com/debanjum/semantic-search && cd semantic-search
#+end_src
** Configure
Configure application search types and their underlying data source/files in ~sample_config.yml~
Use the ~sample_config.yml~ as reference
**** 2. Configure
Add Content Directories for Semantic Search to Docker-Compose
Update [[./docker-compose.yml][docker-compose.yml]] to mount your images, org-mode notes, ledger/beancount directories
If required, edit config settings in [[./docker_sample_config.yml][docker_sample_config.yml]].
** Run
Load ML model, generate embeddings and expose API to query notes, images, transactions etc specified in config YAML
**** 3. Run
#+begin_src shell
docker-compose up -d
#+end_src
#+begin_src shell
python3 -m src.main -c=sample_config.yml -vv
#+end_src
*** Setup on Local Machine
**** 1. Install Dependencies
1. Install Python3 [Required[
2. [[https://docs.conda.io/projects/conda/en/latest/user-guide/install/index.html][Install Conda]] [Required]
3. Install Exiftool [Optional]
#+begin_src shell
sudo apt-get -y install libimage-exiftool-perl
#+end_src
**** 2. Install Semantic Search
#+begin_src shell
git clone https://github.com/debanjum/semantic-search && cd semantic-search
conda env create -f environment.yml
conda activate semantic-search
#+end_src
**** 3. Configure
Configure application search types and their underlying data source/files in ~sample_config.yml~
Use the ~sample_config.yml~ as reference
**** 4. Run
Load ML model, generate embeddings and expose API to query notes, images, transactions etc specified in config YAML
#+begin_src shell
python3 -m src.main -c=sample_config.yml -vv
#+end_src
** Use
- *Semantic Search via Emacs*
@ -39,19 +59,29 @@
- *Semantic Search via API*
- Query: ~GET~ [[http://localhost:8000/search?q=%22what%20is%20the%20meaning%20of%20life%22][http://localhost:8000/search?q="What is the meaning of life"&t=notes]]
- Regenerate Embeddings: ~GET~ [[http://localhost:8000/regenerate][http://localhost:8000/regenerate?t=image]]
- Regenerate Embeddings: ~GET~ [[http://localhost:8000/regenerate][http://localhost:8000/regenerate]]
- [[http://localhost:8000/docs][Semantic Search API Docs]]
- *UI to Edit Config*
- [[https://localhost:8000/ui][Config UI]]
** Upgrade
#+begin_src shell
cd semantic-search
git pull origin master
conda env update -f environment.yml
conda activate semantic-search
#+end_src
*** Using Docker
#+begin_src shell
docker-compose up
#+end_src
*** On Local Machine
#+begin_src shell
cd semantic-search
git pull origin master
conda env update -f environment.yml
conda activate semantic-search
#+end_src
** Acknowledgments
- [[https://huggingface.co/sentence-transformers/msmarco-MiniLM-L-6-v3][MiniLM Model]] for Asymmetric Text Search. See [[https://www.sbert.net/examples/applications/retrieve_rerank/README.html][SBert Documentation]]
- [[https://github.com/openai/CLIP][OpenAI CLIP Model]] for Image Search. See [[https://www.sbert.net/examples/applications/image-search/README.html][SBert Documentation]]
- Charles Cave for [[http://members.optusnet.com.au/~charles57/GTD/orgnode.html][OrgNode Parser]]
- Sven Marnach for [[https://github.com/smarnach/pyexiftool/blob/master/exiftool.py][PyExifTool]]
- Sven Marnach for [[https://github.com/smarnach/pyexiftool/blob/master/exiftool.py][PyExifTool]]

32
docker-compose.yml Normal file
View file

@ -0,0 +1,32 @@
version: "3.9"
services:
server:
build:
context: .
dockerfile: Dockerfile
args:
- PORT=8000
ports:
# If changing the local port (left hand side), no other changes required.
# If changing the remote port (right hand side),
# change the port in the args in the build section,
# as well as the port in the command section to match
- "8000:8000"
working_dir: /app
volumes:
- .:/app
# These mounted volumes hold the raw data that should be indexed for search.
# The path in your local directory (left hand side)
# points to the files you want to index.
# The path of the mounted directory (right hand side),
# must match the path prefix in your config file.
- ./tests/data/:/data/notes/
- ./tests/data/:/data/images/
- ./tests/data/:/data/ledger/
- ./tests/data/:/data/music/
# It's ok if you don't have existing embeddings.
# You can set this volume to point to an empty folder.
- ./tests/data/:/data/generated/
# Use 0.0.0.0 to explicitly set the host ip for the service on the container. https://pythonspeed.com/articles/docker-connection-refused/
command: --host="0.0.0.0" --port=8000 -c=docker_sample_config.yml -vv

47
docker_sample_config.yml Normal file
View file

@ -0,0 +1,47 @@
content-type:
# The /data/folder/ prefix to the folders is here because this is
# the directory to which the local files are copied in the docker-compose.
# If changing, the docker-compose volumes should also be changed to match.
org:
input-files: null
input-filter: "/data/notes/*.org"
compressed-jsonl: "/data/generated/.notes.json.gz"
embeddings-file: "/data/generated/.note_embeddings.pt"
ledger:
input-files: null
input-filter: /data/ledger/*.beancount
compressed-jsonl: /data/generated/.transactions.jsonl.gz
embeddings-file: /data/generated/.transaction_embeddings.pt
image:
input-directory: "/data/images/"
embeddings-file: "/data/generated/.image_embeddings.pt"
batch-size: 50
use-xmp-metadata: true
music:
input-files: null
input-filter: "/data/music/*.org"
compressed-jsonl: "/data/generated/.songs.jsonl.gz"
embeddings-file: "/data/generated/.song_embeddings.pt"
search-type:
symmetric:
encoder: "sentence-transformers/paraphrase-MiniLM-L6-v2"
cross-encoder: "cross-encoder/ms-marco-MiniLM-L-6-v2"
model_directory: "/data/models/.symmetric"
asymmetric:
encoder: "sentence-transformers/msmarco-MiniLM-L-6-v3"
cross-encoder: "cross-encoder/ms-marco-MiniLM-L-6-v2"
model_directory: "/data/models/.asymmetric"
image:
encoder: "clip-ViT-B-32"
model_directory: "/data/models/.image_encoder"
processor:
conversation:
openai-api-key: null
conversation-logfile: "/data/conversation/.conversation_logs.json"

View file

@ -15,7 +15,7 @@ content-type:
input-directory: "tests/data"
embeddings-file: "tests/data/.image_embeddings.pt"
batch-size: 50
use-xmp-metadata: "no"
use-xmp-metadata: false
music:
input-files: ["tests/data/music.org"]

View file

@ -20,7 +20,7 @@ class TextContentConfig(ConfigBase):
embeddings_file: Optional[Path]
class ImageContentConfig(ConfigBase):
use_xmp_metadata: Optional[str]
use_xmp_metadata: Optional[bool]
batch_size: Optional[int]
input_directory: Optional[Path]
input_filter: Optional[str]