Listening for new data

note

In SuperDuperDB, AI models may be configured to listen for newly inserted data. Outputs will be computed over that data and saved back to the data-backend.

In this example we show how to configure 3 models to interact when new data is added.

A featurizing computer vision model (images -> vectors).
2 models evaluating image-2-text similarity to a set of key-words.

We use an open-source model "CLIP" which we install via pip directly from GitHub. You can read more about installing requirements on our docs here.

!pip install git+https://github.com/openai/CLIP.git

Outputs

Collecting git+https://github.com/openai/CLIP.git
Cloning https://github.com/openai/CLIP.git to /private/var/folders/3h/p6qzszds1c7gtbmt_2qq0tvm0000gn/T/pip-req-build-spx4v54y
Running command git clone --filter=blob:none --quiet https://github.com/openai/CLIP.git /private/var/folders/3h/p6qzszds1c7gtbmt_2qq0tvm0000gn/T/pip-req-build-spx4v54y
Resolved https://github.com/openai/CLIP.git to commit a1d071733d7111c9c014f024669f959182114e33
Installing build dependencies ... [?25ldone
[?25h  Getting requirements to build wheel ... [?25ldone
[?25h  Preparing metadata (pyproject.toml) ... [?25ldone
[?25hRequirement already satisfied: ftfy in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from clip==1.0) (6.2.0)
Requirement already satisfied: regex in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from clip==1.0) (2023.12.25)
Requirement already satisfied: tqdm in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from clip==1.0) (4.66.2)
Requirement already satisfied: torch in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from clip==1.0) (2.2.2)
Requirement already satisfied: torchvision in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from clip==1.0) (0.17.2)
Requirement already satisfied: wcwidth<0.3.0,>=0.2.12 in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from ftfy->clip==1.0) (0.2.13)
Requirement already satisfied: filelock in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from torch->clip==1.0) (3.13.3)
Requirement already satisfied: typing-extensions>=4.8.0 in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from torch->clip==1.0) (4.11.0)
Requirement already satisfied: sympy in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from torch->clip==1.0) (1.12)
Requirement already satisfied: networkx in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from torch->clip==1.0) (3.3)
Requirement already satisfied: jinja2 in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from torch->clip==1.0) (3.1.3)
Requirement already satisfied: fsspec in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from torch->clip==1.0) (2024.2.0)
Requirement already satisfied: numpy in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from torchvision->clip==1.0) (1.26.4)
Requirement already satisfied: pillow!=8.3.*,>=5.3.0 in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from torchvision->clip==1.0) (10.3.0)
Requirement already satisfied: MarkupSafe>=2.0 in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from jinja2->torch->clip==1.0) (2.1.5)
Requirement already satisfied: mpmath>=0.19 in /Users/dodo/.pyenv/versions/3.11.7/envs/superduperdb-3.11/lib/python3.11/site-packages (from sympy->torch->clip==1.0) (1.3.0)
[1m[[0m[34;49mnotice[0m[1;39;49m][0m[39;49m A new release of pip is available: [0m[31;49m23.2.1[0m[39;49m -> [0m[32;49m24.0[0m
[1m[[0m[34;49mnotice[0m[1;39;49m][0m[39;49m To update, run: [0m[32;49mpip install --upgrade pip[0m

We apply our setup to images from the cats and dogs dataset. We've prepared a subset especially for quick experimentation.

!curl -O https://superduperdb-public-demo.s3.amazonaws.com/images.zip && unzip images.zip
from PIL import Image
import os

data = [f'images/{x}' for x in os.listdir('./images') if x.endswith('png')]
data = [{'img': Image.open(path)} for path in data]

Outputs

Now that we've prepared these records we can insert this data "directly" into the database with a standard MongoDB insert statement. (Notice however the difference from pymongo with the .execute() call.)

from superduperdb import superduper, Document

db = superduper('mongomock://')

db['images'].insert_many([Document(r) for r in data[:-1]]).execute()

Outputs

2024-May-27 11:33:18.45| INFO     | Duncans-MBP.fritz.box| superduperdb.base.build:69   | Data Client is ready. mongomock.MongoClient('localhost', 27017)
2024-May-27 11:33:18.47| INFO     | Duncans-MBP.fritz.box| superduperdb.base.build:42   | Connecting to Metadata Client with engine:  mongomock.MongoClient('localhost', 27017)
2024-May-27 11:33:18.47| INFO     | Duncans-MBP.fritz.box| superduperdb.base.build:155  | Connecting to compute client: None
2024-May-27 11:33:18.47| INFO     | Duncans-MBP.fritz.box| superduperdb.base.datalayer:85   | Building Data Layer
2024-May-27 11:33:18.47| INFO     | Duncans-MBP.fritz.box| superduperdb.base.build:220  | Configuration:
+---------------+--------------+
| Configuration |    Value     |
+---------------+--------------+
|  Data Backend | mongomock:// |
+---------------+--------------+
2024-May-27 11:33:18.49| WARNING  | Duncans-MBP.fritz.box| superduperdb.misc.annotations:117  | add is deprecated and will be removed in a future release.
2024-May-27 11:33:18.49| WARNING  | Duncans-MBP.fritz.box| superduperdb.backends.local.artifacts:82   | File /tmp/e6eb888f0b0fbbab905029cb309537b9383919a6 already exists
2024-May-27 11:33:18.49| WARNING  | Duncans-MBP.fritz.box| superduperdb.backends.local.artifacts:82   | File /tmp/ee1a946181f065af29a3c8637b2858b153d8fc8e already exists
2024-May-27 11:34:10.59| INFO     | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:37   | Submitting job. function:<function callable_job at 0x10cd4be20>
2024-May-27 11:34:10.99| SUCCESS  | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:43   | Job submitted on <superduperdb.backends.local.compute.LocalComputeBackend object at 0x13ecc76d0>.  function:<function callable_job at 0x10cd4be20> future:cc9d945c-c1d2-474f-a869-a96dac41a32a

([ObjectId('665453924260c60bfe3e8084'),
ObjectId('665453924260c60bfe3e8085'),
ObjectId('665453924260c60bfe3e8086'),
ObjectId('665453924260c60bfe3e8087'),
ObjectId('665453924260c60bfe3e8088'),
ObjectId('665453924260c60bfe3e8089'),
ObjectId('665453924260c60bfe3e808a'),
ObjectId('665453924260c60bfe3e808b'),
ObjectId('665453924260c60bfe3e808c'),
ObjectId('665453924260c60bfe3e808d'),
ObjectId('665453924260c60bfe3e808e'),
ObjectId('665453924260c60bfe3e808f'),
ObjectId('665453924260c60bfe3e8090'),
ObjectId('665453924260c60bfe3e8091'),
ObjectId('665453924260c60bfe3e8092'),
ObjectId('665453924260c60bfe3e8093'),
ObjectId('665453924260c60bfe3e8094'),
ObjectId('665453924260c60bfe3e8095'),
ObjectId('665453924260c60bfe3e8096'),
ObjectId('665453924260c60bfe3e8097'),
ObjectId('665453924260c60bfe3e8098'),
ObjectId('665453924260c60bfe3e8099'),
ObjectId('665453924260c60bfe3e809a'),
ObjectId('665453924260c60bfe3e809b'),
ObjectId('665453924260c60bfe3e809c'),
ObjectId('665453924260c60bfe3e809d'),
ObjectId('665453924260c60bfe3e809e'),
ObjectId('665453924260c60bfe3e809f'),
ObjectId('665453924260c60bfe3e80a0'),
ObjectId('665453924260c60bfe3e80a1'),
ObjectId('665453924260c60bfe3e80a2'),
ObjectId('665453924260c60bfe3e80a3'),
ObjectId('665453924260c60bfe3e80a4'),
ObjectId('665453924260c60bfe3e80a5'),
ObjectId('665453924260c60bfe3e80a6'),
ObjectId('665453924260c60bfe3e80a7'),
ObjectId('665453924260c60bfe3e80a8'),
ObjectId('665453924260c60bfe3e80a9'),
ObjectId('665453924260c60bfe3e80aa'),
ObjectId('665453924260c60bfe3e80ab'),
ObjectId('665453924260c60bfe3e80ac'),
ObjectId('665453924260c60bfe3e80ad'),
ObjectId('665453924260c60bfe3e80ae'),
ObjectId('665453924260c60bfe3e80af'),
ObjectId('665453924260c60bfe3e80b0'),
ObjectId('665453924260c60bfe3e80b1'),
ObjectId('665453924260c60bfe3e80b2'),
ObjectId('665453924260c60bfe3e80b3'),
ObjectId('665453924260c60bfe3e80b4'),
ObjectId('665453924260c60bfe3e80b5'),
ObjectId('665453924260c60bfe3e80b6'),
ObjectId('665453924260c60bfe3e80b7'),
ObjectId('665453924260c60bfe3e80b8'),
ObjectId('665453924260c60bfe3e80b9'),
ObjectId('665453924260c60bfe3e80ba'),
ObjectId('665453924260c60bfe3e80bb'),
ObjectId('665453924260c60bfe3e80bc'),
ObjectId('665453924260c60bfe3e80bd'),
ObjectId('665453924260c60bfe3e80be'),
ObjectId('665453924260c60bfe3e80bf'),
ObjectId('665453924260c60bfe3e80c0'),
ObjectId('665453924260c60bfe3e80c1'),
ObjectId('665453924260c60bfe3e80c2'),
ObjectId('665453924260c60bfe3e80c3'),
ObjectId('665453924260c60bfe3e80c4'),
ObjectId('665453924260c60bfe3e80c5'),
ObjectId('665453924260c60bfe3e80c6'),
ObjectId('665453924260c60bfe3e80c7'),
ObjectId('665453924260c60bfe3e80c8'),
ObjectId('665453924260c60bfe3e80c9'),
ObjectId('665453924260c60bfe3e80ca'),
ObjectId('665453924260c60bfe3e80cb'),
ObjectId('665453924260c60bfe3e80cc'),
ObjectId('665453924260c60bfe3e80cd'),
ObjectId('665453924260c60bfe3e80ce'),
ObjectId('665453924260c60bfe3e80cf'),
ObjectId('665453924260c60bfe3e80d0'),
ObjectId('665453924260c60bfe3e80d1'),
ObjectId('665453924260c60bfe3e80d2'),
ObjectId('665453924260c60bfe3e80d3'),
ObjectId('665453924260c60bfe3e80d4'),
ObjectId('665453924260c60bfe3e80d5'),
ObjectId('665453924260c60bfe3e80d6'),
ObjectId('665453924260c60bfe3e80d7'),
ObjectId('665453924260c60bfe3e80d8'),
ObjectId('665453924260c60bfe3e80d9'),
ObjectId('665453924260c60bfe3e80da'),
ObjectId('665453924260c60bfe3e80db'),
ObjectId('665453924260c60bfe3e80dc'),
ObjectId('665453924260c60bfe3e80dd'),
ObjectId('665453924260c60bfe3e80de'),
ObjectId('665453924260c60bfe3e80df'),
ObjectId('665453924260c60bfe3e80e0'),
ObjectId('665453924260c60bfe3e80e1'),
ObjectId('665453924260c60bfe3e80e2'),
ObjectId('665453924260c60bfe3e80e3'),
ObjectId('665453924260c60bfe3e80e4'),
ObjectId('665453924260c60bfe3e80e5'),
ObjectId('665453924260c60bfe3e80e6'),
ObjectId('665453924260c60bfe3e80e7'),
ObjectId('665453924260c60bfe3e80e8'),
ObjectId('665453924260c60bfe3e80e9'),
ObjectId('665453924260c60bfe3e80ea'),
ObjectId('665453924260c60bfe3e80eb'),
ObjectId('665453924260c60bfe3e80ec'),
ObjectId('665453924260c60bfe3e80ed'),
ObjectId('665453924260c60bfe3e80ee'),
ObjectId('665453924260c60bfe3e80ef'),
ObjectId('665453924260c60bfe3e80f0'),
ObjectId('665453924260c60bfe3e80f1'),
ObjectId('665453924260c60bfe3e80f2'),
ObjectId('665453924260c60bfe3e80f3'),
ObjectId('665453924260c60bfe3e80f4'),
ObjectId('665453924260c60bfe3e80f5'),
ObjectId('665453924260c60bfe3e80f6'),
ObjectId('665453924260c60bfe3e80f7'),
ObjectId('665453924260c60bfe3e80f8'),
ObjectId('665453924260c60bfe3e80f9'),
ObjectId('665453924260c60bfe3e80fa'),
ObjectId('665453924260c60bfe3e80fb'),
ObjectId('665453924260c60bfe3e80fc'),
ObjectId('665453924260c60bfe3e80fd'),
ObjectId('665453924260c60bfe3e80fe'),
ObjectId('665453924260c60bfe3e80ff'),
ObjectId('665453924260c60bfe3e8100'),
ObjectId('665453924260c60bfe3e8101'),
ObjectId('665453924260c60bfe3e8102'),
ObjectId('665453924260c60bfe3e8103'),
ObjectId('665453924260c60bfe3e8104'),
ObjectId('665453924260c60bfe3e8105'),
ObjectId('665453924260c60bfe3e8106'),
ObjectId('665453924260c60bfe3e8107'),
ObjectId('665453924260c60bfe3e8108'),
ObjectId('665453924260c60bfe3e8109'),
ObjectId('665453924260c60bfe3e810a'),
ObjectId('665453924260c60bfe3e810b'),
ObjectId('665453924260c60bfe3e810c'),
ObjectId('665453924260c60bfe3e810d'),
ObjectId('665453924260c60bfe3e810e'),
ObjectId('665453924260c60bfe3e810f'),
ObjectId('665453924260c60bfe3e8110'),
ObjectId('665453924260c60bfe3e8111'),
ObjectId('665453924260c60bfe3e8112'),
ObjectId('665453924260c60bfe3e8113'),
ObjectId('665453924260c60bfe3e8114'),
ObjectId('665453924260c60bfe3e8115'),
ObjectId('665453924260c60bfe3e8116'),
ObjectId('665453924260c60bfe3e8117'),
ObjectId('665453924260c60bfe3e8118'),
ObjectId('665453924260c60bfe3e8119'),
ObjectId('665453924260c60bfe3e811a'),
ObjectId('665453924260c60bfe3e811b'),
ObjectId('665453924260c60bfe3e811c'),
ObjectId('665453924260c60bfe3e811d'),
ObjectId('665453924260c60bfe3e811e'),
ObjectId('665453924260c60bfe3e811f'),
ObjectId('665453924260c60bfe3e8120'),
ObjectId('665453924260c60bfe3e8121'),
ObjectId('665453924260c60bfe3e8122'),
ObjectId('665453924260c60bfe3e8123'),
ObjectId('665453924260c60bfe3e8124'),
ObjectId('665453924260c60bfe3e8125'),
ObjectId('665453924260c60bfe3e8126'),
ObjectId('665453924260c60bfe3e8127'),
ObjectId('665453924260c60bfe3e8128'),
ObjectId('665453924260c60bfe3e8129'),
ObjectId('665453924260c60bfe3e812a'),
ObjectId('665453924260c60bfe3e812b'),
ObjectId('665453924260c60bfe3e812c'),
ObjectId('665453924260c60bfe3e812d'),
ObjectId('665453924260c60bfe3e812e'),
ObjectId('665453924260c60bfe3e812f'),
ObjectId('665453924260c60bfe3e8130'),
ObjectId('665453924260c60bfe3e8131'),
ObjectId('665453924260c60bfe3e8132'),
ObjectId('665453924260c60bfe3e8133'),
ObjectId('665453924260c60bfe3e8134'),
ObjectId('665453924260c60bfe3e8135'),
ObjectId('665453924260c60bfe3e8136'),
ObjectId('665453924260c60bfe3e8137'),
ObjectId('665453924260c60bfe3e8138'),
ObjectId('665453924260c60bfe3e8139'),
ObjectId('665453924260c60bfe3e813a'),
ObjectId('665453924260c60bfe3e813b'),
ObjectId('665453924260c60bfe3e813c'),
ObjectId('665453924260c60bfe3e813d'),
ObjectId('665453924260c60bfe3e813e'),
ObjectId('665453924260c60bfe3e813f'),
ObjectId('665453924260c60bfe3e8140'),
ObjectId('665453924260c60bfe3e8141'),
ObjectId('665453924260c60bfe3e8142'),
ObjectId('665453924260c60bfe3e8143'),
ObjectId('665453924260c60bfe3e8144'),
ObjectId('665453924260c60bfe3e8145'),
ObjectId('665453924260c60bfe3e8146'),
ObjectId('665453924260c60bfe3e8147'),
ObjectId('665453924260c60bfe3e8148'),
ObjectId('665453924260c60bfe3e8149'),
ObjectId('665453924260c60bfe3e814a'),
ObjectId('665453924260c60bfe3e814b'),
ObjectId('665453924260c60bfe3e814c'),
ObjectId('665453924260c60bfe3e814d'),
ObjectId('665453924260c60bfe3e814e'),
ObjectId('665453924260c60bfe3e814f'),
ObjectId('665453924260c60bfe3e8150'),
ObjectId('665453924260c60bfe3e8151'),
ObjectId('665453924260c60bfe3e8152'),
ObjectId('665453924260c60bfe3e8153'),
ObjectId('665453924260c60bfe3e8154'),
ObjectId('665453924260c60bfe3e8155'),
ObjectId('665453924260c60bfe3e8156'),
ObjectId('665453924260c60bfe3e8157'),
ObjectId('665453924260c60bfe3e8158'),
ObjectId('665453924260c60bfe3e8159'),
ObjectId('665453924260c60bfe3e815a'),
ObjectId('665453924260c60bfe3e815b'),
ObjectId('665453924260c60bfe3e815c'),
ObjectId('665453924260c60bfe3e815d'),
ObjectId('665453924260c60bfe3e815e'),
ObjectId('665453924260c60bfe3e815f'),
ObjectId('665453924260c60bfe3e8160'),
ObjectId('665453924260c60bfe3e8161'),
ObjectId('665453924260c60bfe3e8162'),
ObjectId('665453924260c60bfe3e8163'),
ObjectId('665453924260c60bfe3e8164'),
ObjectId('665453924260c60bfe3e8165'),
ObjectId('665453924260c60bfe3e8166'),
ObjectId('665453924260c60bfe3e8167'),
ObjectId('665453924260c60bfe3e8168'),
ObjectId('665453924260c60bfe3e8169'),
ObjectId('665453924260c60bfe3e816a'),
ObjectId('665453924260c60bfe3e816b'),
ObjectId('665453924260c60bfe3e816c'),
ObjectId('665453924260c60bfe3e816d'),
ObjectId('665453924260c60bfe3e816e'),
ObjectId('665453924260c60bfe3e816f'),
ObjectId('665453924260c60bfe3e8170'),
ObjectId('665453924260c60bfe3e8171'),
ObjectId('665453924260c60bfe3e8172'),
ObjectId('665453924260c60bfe3e8173'),
ObjectId('665453924260c60bfe3e8174'),
ObjectId('665453924260c60bfe3e8175'),
ObjectId('665453924260c60bfe3e8176'),
ObjectId('665453924260c60bfe3e8177'),
ObjectId('665453924260c60bfe3e8178'),
ObjectId('665453924260c60bfe3e8179'),
ObjectId('665453924260c60bfe3e817a'),
ObjectId('665453924260c60bfe3e817b'),
ObjectId('665453924260c60bfe3e817c'),
ObjectId('665453924260c60bfe3e817d'),
ObjectId('665453924260c60bfe3e817e'),
ObjectId('665453924260c60bfe3e817f'),
ObjectId('665453924260c60bfe3e8180'),
ObjectId('665453924260c60bfe3e8181'),
ObjectId('665453924260c60bfe3e8182'),
ObjectId('665453924260c60bfe3e8183'),
ObjectId('665453924260c60bfe3e8184'),
ObjectId('665453924260c60bfe3e8185'),
ObjectId('665453924260c60bfe3e8186'),
ObjectId('665453924260c60bfe3e8187'),
ObjectId('665453924260c60bfe3e8188'),
ObjectId('665453924260c60bfe3e8189'),
ObjectId('665453924260c60bfe3e818a'),
ObjectId('665453924260c60bfe3e818b'),
ObjectId('665453924260c60bfe3e818c'),
ObjectId('665453924260c60bfe3e818d'),
ObjectId('665453924260c60bfe3e818e'),
ObjectId('665453924260c60bfe3e818f'),
ObjectId('665453924260c60bfe3e8190'),
ObjectId('665453924260c60bfe3e8191'),
ObjectId('665453924260c60bfe3e8192'),
ObjectId('665453924260c60bfe3e8193'),
ObjectId('665453924260c60bfe3e8194'),
ObjectId('665453924260c60bfe3e8195'),
ObjectId('665453924260c60bfe3e8196'),
ObjectId('665453924260c60bfe3e8197'),
ObjectId('665453924260c60bfe3e8198'),
ObjectId('665453924260c60bfe3e8199'),
ObjectId('665453924260c60bfe3e819a'),
ObjectId('665453924260c60bfe3e819b'),
ObjectId('665453924260c60bfe3e819c'),
ObjectId('665453924260c60bfe3e819d'),
ObjectId('665453924260c60bfe3e819e'),
ObjectId('665453924260c60bfe3e819f'),
ObjectId('665453924260c60bfe3e81a0'),
ObjectId('665453924260c60bfe3e81a1'),
ObjectId('665453924260c60bfe3e81a2'),
ObjectId('665453924260c60bfe3e81a3'),
ObjectId('665453924260c60bfe3e81a4'),
ObjectId('665453924260c60bfe3e81a5'),
ObjectId('665453924260c60bfe3e81a6'),
ObjectId('665453924260c60bfe3e81a7'),
ObjectId('665453924260c60bfe3e81a8'),
ObjectId('665453924260c60bfe3e81a9'),
ObjectId('665453924260c60bfe3e81aa'),
ObjectId('665453924260c60bfe3e81ab'),
ObjectId('665453924260c60bfe3e81ac'),
ObjectId('665453924260c60bfe3e81ad'),
ObjectId('665453924260c60bfe3e81ae'),
ObjectId('665453924260c60bfe3e81af'),
ObjectId('665453924260c60bfe3e81b0'),
ObjectId('665453924260c60bfe3e81b1'),
ObjectId('665453924260c60bfe3e81b2'),
ObjectId('665453924260c60bfe3e81b3'),
ObjectId('665453924260c60bfe3e81b4'),
ObjectId('665453924260c60bfe3e81b5'),
ObjectId('665453924260c60bfe3e81b6'),
ObjectId('665453924260c60bfe3e81b7'),
ObjectId('665453924260c60bfe3e81b8'),
ObjectId('665453924260c60bfe3e81b9'),
ObjectId('665453924260c60bfe3e81ba'),
ObjectId('665453924260c60bfe3e81bb'),
ObjectId('665453924260c60bfe3e81bc'),
ObjectId('665453924260c60bfe3e81bd'),
ObjectId('665453924260c60bfe3e81be'),
ObjectId('665453924260c60bfe3e81bf'),
ObjectId('665453924260c60bfe3e81c0'),
ObjectId('665453924260c60bfe3e81c1'),
ObjectId('665453924260c60bfe3e81c2'),
ObjectId('665453924260c60bfe3e81c3'),
ObjectId('665453924260c60bfe3e81c4'),
ObjectId('665453924260c60bfe3e81c5'),
ObjectId('665453924260c60bfe3e81c6'),
ObjectId('665453924260c60bfe3e81c7'),
ObjectId('665453924260c60bfe3e81c8'),
ObjectId('665453924260c60bfe3e81c9'),
ObjectId('665453924260c60bfe3e81ca'),
ObjectId('665453924260c60bfe3e81cb'),
ObjectId('665453924260c60bfe3e81cc'),
ObjectId('665453924260c60bfe3e81cd'),
ObjectId('665453924260c60bfe3e81ce'),
ObjectId('665453924260c60bfe3e81cf'),
ObjectId('665453924260c60bfe3e81d0'),
ObjectId('665453924260c60bfe3e81d1'),
ObjectId('665453924260c60bfe3e81d2'),
ObjectId('665453924260c60bfe3e81d3'),
ObjectId('665453924260c60bfe3e81d4'),
ObjectId('665453924260c60bfe3e81d5'),
ObjectId('665453924260c60bfe3e81d6'),
ObjectId('665453924260c60bfe3e81d7'),
ObjectId('665453924260c60bfe3e81d8'),
ObjectId('665453924260c60bfe3e81d9'),
ObjectId('665453924260c60bfe3e81da'),
ObjectId('665453924260c60bfe3e81db'),
ObjectId('665453924260c60bfe3e81dc'),
ObjectId('665453924260c60bfe3e81dd'),
ObjectId('665453924260c60bfe3e81de'),
ObjectId('665453924260c60bfe3e81df'),
ObjectId('665453924260c60bfe3e81e0'),
ObjectId('665453924260c60bfe3e81e1'),
ObjectId('665453924260c60bfe3e81e2'),
ObjectId('665453924260c60bfe3e81e3'),
ObjectId('665453924260c60bfe3e81e4'),
ObjectId('665453924260c60bfe3e81e5'),
ObjectId('665453924260c60bfe3e81e6'),
ObjectId('665453924260c60bfe3e81e7'),
ObjectId('665453924260c60bfe3e81e8'),
ObjectId('665453924260c60bfe3e81e9'),
ObjectId('665453924260c60bfe3e81ea'),
ObjectId('665453924260c60bfe3e81eb'),
ObjectId('665453924260c60bfe3e81ec'),
ObjectId('665453924260c60bfe3e81ed'),
ObjectId('665453924260c60bfe3e81ee'),
ObjectId('665453924260c60bfe3e81ef'),
ObjectId('665453924260c60bfe3e81f0'),
ObjectId('665453924260c60bfe3e81f1'),
ObjectId('665453924260c60bfe3e81f2'),
ObjectId('665453924260c60bfe3e81f3'),
ObjectId('665453924260c60bfe3e81f4'),
ObjectId('665453924260c60bfe3e81f5'),
ObjectId('665453924260c60bfe3e81f6'),
ObjectId('665453924260c60bfe3e81f7'),
ObjectId('665453924260c60bfe3e81f8'),
ObjectId('665453924260c60bfe3e81f9'),
ObjectId('665453924260c60bfe3e81fa'),
ObjectId('665453924260c60bfe3e81fb'),
ObjectId('665453924260c60bfe3e81fc'),
ObjectId('665453924260c60bfe3e81fd'),
ObjectId('665453924260c60bfe3e81fe'),
ObjectId('665453924260c60bfe3e81ff'),
ObjectId('665453924260c60bfe3e8200'),
ObjectId('665453924260c60bfe3e8201'),
ObjectId('665453924260c60bfe3e8202'),
ObjectId('665453924260c60bfe3e8203'),
ObjectId('665453924260c60bfe3e8204'),
ObjectId('665453924260c60bfe3e8205'),
ObjectId('665453924260c60bfe3e8206'),
ObjectId('665453924260c60bfe3e8207'),
ObjectId('665453924260c60bfe3e8208'),
ObjectId('665453924260c60bfe3e8209'),
ObjectId('665453924260c60bfe3e820a'),
ObjectId('665453924260c60bfe3e820b'),
ObjectId('665453924260c60bfe3e820c'),
ObjectId('665453924260c60bfe3e820d'),
ObjectId('665453924260c60bfe3e820e'),
ObjectId('665453924260c60bfe3e820f'),
ObjectId('665453924260c60bfe3e8210'),
ObjectId('665453924260c60bfe3e8211'),
ObjectId('665453924260c60bfe3e8212'),
ObjectId('665453924260c60bfe3e8213'),
ObjectId('665453924260c60bfe3e8214'),
ObjectId('665453924260c60bfe3e8215'),
ObjectId('665453924260c60bfe3e8216'),
ObjectId('665453924260c60bfe3e8217'),
ObjectId('665453924260c60bfe3e8218'),
ObjectId('665453924260c60bfe3e8219'),
ObjectId('665453924260c60bfe3e821a'),
ObjectId('665453924260c60bfe3e821b'),
ObjectId('665453924260c60bfe3e821c'),
ObjectId('665453924260c60bfe3e821d'),
ObjectId('665453924260c60bfe3e821e'),
ObjectId('665453924260c60bfe3e821f'),
ObjectId('665453924260c60bfe3e8220'),
ObjectId('665453924260c60bfe3e8221'),
ObjectId('665453924260c60bfe3e8222'),
ObjectId('665453924260c60bfe3e8223'),
ObjectId('665453924260c60bfe3e8224'),
ObjectId('665453924260c60bfe3e8225'),
ObjectId('665453924260c60bfe3e8226'),
ObjectId('665453924260c60bfe3e8227'),
ObjectId('665453924260c60bfe3e8228'),
ObjectId('665453924260c60bfe3e8229'),
ObjectId('665453924260c60bfe3e822a'),
ObjectId('665453924260c60bfe3e822b'),
ObjectId('665453924260c60bfe3e822c'),
ObjectId('665453924260c60bfe3e822d'),
ObjectId('665453924260c60bfe3e822e'),
ObjectId('665453924260c60bfe3e822f'),
ObjectId('665453924260c60bfe3e8230'),
ObjectId('665453924260c60bfe3e8231'),
ObjectId('665453924260c60bfe3e8232'),
ObjectId('665453924260c60bfe3e8233'),
ObjectId('665453924260c60bfe3e8234'),
ObjectId('665453924260c60bfe3e8235'),
ObjectId('665453924260c60bfe3e8236'),
ObjectId('665453924260c60bfe3e8237'),
ObjectId('665453924260c60bfe3e8238'),
ObjectId('665453924260c60bfe3e8239'),
ObjectId('665453924260c60bfe3e823a'),
ObjectId('665453924260c60bfe3e823b'),
ObjectId('665453924260c60bfe3e823c'),
ObjectId('665453924260c60bfe3e823d'),
ObjectId('665453924260c60bfe3e823e'),
ObjectId('665453924260c60bfe3e823f'),
ObjectId('665453924260c60bfe3e8240'),
ObjectId('665453924260c60bfe3e8241'),
ObjectId('665453924260c60bfe3e8242'),
ObjectId('665453924260c60bfe3e8243'),
ObjectId('665453924260c60bfe3e8244'),
ObjectId('665453924260c60bfe3e8245'),
ObjectId('665453924260c60bfe3e8246'),
ObjectId('665453924260c60bfe3e8247'),
ObjectId('665453924260c60bfe3e8248'),
ObjectId('665453924260c60bfe3e8249'),
ObjectId('665453924260c60bfe3e824a'),
ObjectId('665453924260c60bfe3e824b'),
ObjectId('665453924260c60bfe3e824c'),
ObjectId('665453924260c60bfe3e824d'),
ObjectId('665453924260c60bfe3e824e'),
ObjectId('665453924260c60bfe3e824f'),
ObjectId('665453924260c60bfe3e8250'),
ObjectId('665453924260c60bfe3e8251'),
ObjectId('665453924260c60bfe3e8252'),
ObjectId('665453924260c60bfe3e8253'),
ObjectId('665453924260c60bfe3e8254'),
ObjectId('665453924260c60bfe3e8255'),
ObjectId('665453924260c60bfe3e8256'),
ObjectId('665453924260c60bfe3e8257'),
ObjectId('665453924260c60bfe3e8258'),
ObjectId('665453924260c60bfe3e8259'),
ObjectId('665453924260c60bfe3e825a'),
ObjectId('665453924260c60bfe3e825b'),
ObjectId('665453924260c60bfe3e825c'),
ObjectId('665453924260c60bfe3e825d'),
ObjectId('665453924260c60bfe3e825e'),
ObjectId('665453924260c60bfe3e825f'),
ObjectId('665453924260c60bfe3e8260'),
ObjectId('665453924260c60bfe3e8261'),
ObjectId('665453924260c60bfe3e8262'),
ObjectId('665453924260c60bfe3e8263'),
ObjectId('665453924260c60bfe3e8264'),
ObjectId('665453924260c60bfe3e8265'),
ObjectId('665453924260c60bfe3e8266'),
ObjectId('665453924260c60bfe3e8267'),
ObjectId('665453924260c60bfe3e8268'),
ObjectId('665453924260c60bfe3e8269'),
ObjectId('665453924260c60bfe3e826a'),
ObjectId('665453924260c60bfe3e826b'),
ObjectId('665453924260c60bfe3e826c'),
ObjectId('665453924260c60bfe3e826d'),
ObjectId('665453924260c60bfe3e826e'),
ObjectId('665453924260c60bfe3e826f'),
ObjectId('665453924260c60bfe3e8270'),
ObjectId('665453924260c60bfe3e8271'),
ObjectId('665453924260c60bfe3e8272'),
ObjectId('665453924260c60bfe3e8273'),
ObjectId('665453924260c60bfe3e8274'),
ObjectId('665453924260c60bfe3e8275'),
ObjectId('665453924260c60bfe3e8276'),
ObjectId('665453924260c60bfe3e8277'),
ObjectId('665453924260c60bfe3e8278'),
ObjectId('665453924260c60bfe3e8279'),
ObjectId('665453924260c60bfe3e827a'),
ObjectId('665453924260c60bfe3e827b'),
ObjectId('665453924260c60bfe3e827c'),
ObjectId('665453924260c60bfe3e827d'),
ObjectId('665453924260c60bfe3e827e'),
ObjectId('665453924260c60bfe3e827f'),
ObjectId('665453924260c60bfe3e8280'),
ObjectId('665453924260c60bfe3e8281'),
ObjectId('665453924260c60bfe3e8282'),
ObjectId('665453924260c60bfe3e8283'),
ObjectId('665453924260c60bfe3e8284'),
ObjectId('665453924260c60bfe3e8285'),
ObjectId('665453924260c60bfe3e8286'),
ObjectId('665453924260c60bfe3e8287'),
ObjectId('665453924260c60bfe3e8288'),
ObjectId('665453924260c60bfe3e8289'),
ObjectId('665453924260c60bfe3e828a'),
ObjectId('665453924260c60bfe3e828b'),
ObjectId('665453924260c60bfe3e828c'),
ObjectId('665453924260c60bfe3e828d'),
ObjectId('665453924260c60bfe3e828e'),
ObjectId('665453924260c60bfe3e828f'),
ObjectId('665453924260c60bfe3e8290'),
ObjectId('665453924260c60bfe3e8291'),
ObjectId('665453924260c60bfe3e8292'),
ObjectId('665453924260c60bfe3e8293'),
ObjectId('665453924260c60bfe3e8294'),
ObjectId('665453924260c60bfe3e8295'),
ObjectId('665453924260c60bfe3e8296'),
ObjectId('665453924260c60bfe3e8297'),
ObjectId('665453924260c60bfe3e8298'),
ObjectId('665453924260c60bfe3e8299'),
ObjectId('665453924260c60bfe3e829a'),
ObjectId('665453924260c60bfe3e829b'),
ObjectId('665453924260c60bfe3e829c'),
ObjectId('665453924260c60bfe3e829d'),
ObjectId('665453924260c60bfe3e829e'),
ObjectId('665453924260c60bfe3e829f'),
ObjectId('665453924260c60bfe3e82a0'),
ObjectId('665453924260c60bfe3e82a1'),
ObjectId('665453924260c60bfe3e82a2'),
ObjectId('665453924260c60bfe3e82a3'),
ObjectId('665453924260c60bfe3e82a4'),
ObjectId('665453924260c60bfe3e82a5'),
ObjectId('665453924260c60bfe3e82a6'),
ObjectId('665453924260c60bfe3e82a7'),
ObjectId('665453924260c60bfe3e82a8'),
ObjectId('665453924260c60bfe3e82a9'),
ObjectId('665453924260c60bfe3e82aa'),
ObjectId('665453924260c60bfe3e82ab'),
ObjectId('665453924260c60bfe3e82ac'),
ObjectId('665453924260c60bfe3e82ad'),
ObjectId('665453924260c60bfe3e82ae'),
ObjectId('665453924260c60bfe3e82af'),
ObjectId('665453924260c60bfe3e82b0'),
ObjectId('665453924260c60bfe3e82b1'),
ObjectId('665453924260c60bfe3e82b2'),
ObjectId('665453924260c60bfe3e82b3'),
ObjectId('665453924260c60bfe3e82b4'),
ObjectId('665453924260c60bfe3e82b5'),
ObjectId('665453924260c60bfe3e82b6'),
ObjectId('665453924260c60bfe3e82b7'),
ObjectId('665453924260c60bfe3e82b8'),
ObjectId('665453924260c60bfe3e82b9'),
ObjectId('665453924260c60bfe3e82ba'),
ObjectId('665453924260c60bfe3e82bb'),
ObjectId('665453924260c60bfe3e82bc'),
ObjectId('665453924260c60bfe3e82bd'),
ObjectId('665453924260c60bfe3e82be'),
ObjectId('665453924260c60bfe3e82bf'),
ObjectId('665453924260c60bfe3e82c0'),
ObjectId('665453924260c60bfe3e82c1'),
ObjectId('665453924260c60bfe3e82c2'),
ObjectId('665453924260c60bfe3e82c3'),
ObjectId('665453924260c60bfe3e82c4'),
ObjectId('665453924260c60bfe3e82c5'),
ObjectId('665453924260c60bfe3e82c6'),
ObjectId('665453924260c60bfe3e82c7'),
ObjectId('665453924260c60bfe3e82c8'),
ObjectId('665453924260c60bfe3e82c9'),
ObjectId('665453924260c60bfe3e82ca'),
ObjectId('665453924260c60bfe3e82cb'),
ObjectId('665453924260c60bfe3e82cc'),
ObjectId('665453924260c60bfe3e82cd'),
ObjectId('665453924260c60bfe3e82ce'),
ObjectId('665453924260c60bfe3e82cf'),
ObjectId('665453924260c60bfe3e82d0'),
ObjectId('665453924260c60bfe3e82d1'),
ObjectId('665453924260c60bfe3e82d2'),
ObjectId('665453924260c60bfe3e82d3'),
ObjectId('665453924260c60bfe3e82d4'),
ObjectId('665453924260c60bfe3e82d5'),
ObjectId('665453924260c60bfe3e82d6'),
ObjectId('665453924260c60bfe3e82d7'),
ObjectId('665453924260c60bfe3e82d8'),
ObjectId('665453924260c60bfe3e82d9'),
ObjectId('665453924260c60bfe3e82da'),
ObjectId('665453924260c60bfe3e82db'),
ObjectId('665453924260c60bfe3e82dc'),
ObjectId('665453924260c60bfe3e82dd'),
ObjectId('665453924260c60bfe3e82de'),
ObjectId('665453924260c60bfe3e82df'),
ObjectId('665453924260c60bfe3e82e0'),
ObjectId('665453924260c60bfe3e82e1'),
ObjectId('665453924260c60bfe3e82e2'),
ObjectId('665453924260c60bfe3e82e3'),
ObjectId('665453924260c60bfe3e82e4'),
ObjectId('665453924260c60bfe3e82e5'),
ObjectId('665453924260c60bfe3e82e6'),
ObjectId('665453924260c60bfe3e82e7'),
ObjectId('665453924260c60bfe3e82e8'),
ObjectId('665453924260c60bfe3e82e9'),
ObjectId('665453924260c60bfe3e82ea'),
ObjectId('665453924260c60bfe3e82eb'),
ObjectId('665453924260c60bfe3e82ec'),
ObjectId('665453924260c60bfe3e82ed'),
ObjectId('665453924260c60bfe3e82ee'),
ObjectId('665453924260c60bfe3e82ef'),
ObjectId('665453924260c60bfe3e82f0'),
ObjectId('665453924260c60bfe3e82f1'),
ObjectId('665453924260c60bfe3e82f2'),
ObjectId('665453924260c60bfe3e82f3'),
ObjectId('665453924260c60bfe3e82f4'),
ObjectId('665453924260c60bfe3e82f5'),
ObjectId('665453924260c60bfe3e82f6'),
ObjectId('665453924260c60bfe3e82f7'),
ObjectId('665453924260c60bfe3e82f8'),
ObjectId('665453924260c60bfe3e82f9'),
ObjectId('665453924260c60bfe3e82fa'),
ObjectId('665453924260c60bfe3e82fb'),
ObjectId('665453924260c60bfe3e82fc'),
ObjectId('665453924260c60bfe3e82fd'),
ObjectId('665453924260c60bfe3e82fe'),
ObjectId('665453924260c60bfe3e82ff'),
ObjectId('665453924260c60bfe3e8300'),
ObjectId('665453924260c60bfe3e8301'),
ObjectId('665453924260c60bfe3e8302'),
ObjectId('665453924260c60bfe3e8303'),
ObjectId('665453924260c60bfe3e8304'),
ObjectId('665453924260c60bfe3e8305'),
ObjectId('665453924260c60bfe3e8306'),
ObjectId('665453924260c60bfe3e8307'),
ObjectId('665453924260c60bfe3e8308'),
ObjectId('665453924260c60bfe3e8309'),
ObjectId('665453924260c60bfe3e830a'),
ObjectId('665453924260c60bfe3e830b'),
ObjectId('665453924260c60bfe3e830c'),
ObjectId('665453924260c60bfe3e830d'),
ObjectId('665453924260c60bfe3e830e'),
ObjectId('665453924260c60bfe3e830f'),
ObjectId('665453924260c60bfe3e8310'),
ObjectId('665453924260c60bfe3e8311'),
ObjectId('665453924260c60bfe3e8312'),
ObjectId('665453924260c60bfe3e8313'),
ObjectId('665453924260c60bfe3e8314'),
ObjectId('665453924260c60bfe3e8315'),
ObjectId('665453924260c60bfe3e8316'),
ObjectId('665453924260c60bfe3e8317'),
ObjectId('665453924260c60bfe3e8318'),
ObjectId('665453924260c60bfe3e8319'),
ObjectId('665453924260c60bfe3e831a'),
ObjectId('665453924260c60bfe3e831b'),
ObjectId('665453924260c60bfe3e831c'),
ObjectId('665453924260c60bfe3e831d'),
ObjectId('665453924260c60bfe3e831e'),
ObjectId('665453924260c60bfe3e831f'),
ObjectId('665453924260c60bfe3e8320'),
ObjectId('665453924260c60bfe3e8321'),
ObjectId('665453924260c60bfe3e8322'),
ObjectId('665453924260c60bfe3e8323'),
ObjectId('665453924260c60bfe3e8324'),
ObjectId('665453924260c60bfe3e8325'),
ObjectId('665453924260c60bfe3e8326'),
ObjectId('665453924260c60bfe3e8327'),
ObjectId('665453924260c60bfe3e8328'),
ObjectId('665453924260c60bfe3e8329'),
ObjectId('665453924260c60bfe3e832a'),
ObjectId('665453924260c60bfe3e832b'),
ObjectId('665453924260c60bfe3e832c'),
ObjectId('665453924260c60bfe3e832d'),
ObjectId('665453924260c60bfe3e832e'),
ObjectId('665453924260c60bfe3e832f'),
ObjectId('665453924260c60bfe3e8330'),
ObjectId('665453924260c60bfe3e8331'),
ObjectId('665453924260c60bfe3e8332'),
ObjectId('665453924260c60bfe3e8333'),
ObjectId('665453924260c60bfe3e8334'),
ObjectId('665453924260c60bfe3e8335'),
ObjectId('665453924260c60bfe3e8336'),
ObjectId('665453924260c60bfe3e8337'),
ObjectId('665453924260c60bfe3e8338'),
ObjectId('665453924260c60bfe3e8339'),
ObjectId('665453924260c60bfe3e833a'),
ObjectId('665453924260c60bfe3e833b'),
ObjectId('665453924260c60bfe3e833c'),
ObjectId('665453924260c60bfe3e833d'),
ObjectId('665453924260c60bfe3e833e'),
ObjectId('665453924260c60bfe3e833f'),
ObjectId('665453924260c60bfe3e8340'),
ObjectId('665453924260c60bfe3e8341'),
ObjectId('665453924260c60bfe3e8342'),
ObjectId('665453924260c60bfe3e8343'),
ObjectId('665453924260c60bfe3e8344'),
ObjectId('665453924260c60bfe3e8345'),
ObjectId('665453924260c60bfe3e8346'),
ObjectId('665453924260c60bfe3e8347'),
ObjectId('665453924260c60bfe3e8348'),
ObjectId('665453924260c60bfe3e8349'),
ObjectId('665453924260c60bfe3e834a'),
ObjectId('665453924260c60bfe3e834b'),
ObjectId('665453924260c60bfe3e834c'),
ObjectId('665453924260c60bfe3e834d'),
ObjectId('665453924260c60bfe3e834e'),
ObjectId('665453924260c60bfe3e834f'),
ObjectId('665453924260c60bfe3e8350'),
ObjectId('665453924260c60bfe3e8351'),
ObjectId('665453924260c60bfe3e8352'),
ObjectId('665453924260c60bfe3e8353'),
ObjectId('665453924260c60bfe3e8354'),
ObjectId('665453924260c60bfe3e8355'),
ObjectId('665453924260c60bfe3e8356'),
ObjectId('665453924260c60bfe3e8357'),
ObjectId('665453924260c60bfe3e8358'),
ObjectId('665453924260c60bfe3e8359'),
ObjectId('665453924260c60bfe3e835a'),
ObjectId('665453924260c60bfe3e835b'),
ObjectId('665453924260c60bfe3e835c'),
ObjectId('665453924260c60bfe3e835d'),
ObjectId('665453924260c60bfe3e835e'),
ObjectId('665453924260c60bfe3e835f'),
ObjectId('665453924260c60bfe3e8360'),
ObjectId('665453924260c60bfe3e8361'),
ObjectId('665453924260c60bfe3e8362'),
ObjectId('665453924260c60bfe3e8363'),
ObjectId('665453924260c60bfe3e8364'),
ObjectId('665453924260c60bfe3e8365'),
ObjectId('665453924260c60bfe3e8366'),
ObjectId('665453924260c60bfe3e8367'),
ObjectId('665453924260c60bfe3e8368'),
ObjectId('665453924260c60bfe3e8369'),
ObjectId('665453924260c60bfe3e836a'),
ObjectId('665453924260c60bfe3e836b'),
ObjectId('665453924260c60bfe3e836c'),
ObjectId('665453924260c60bfe3e836d'),
ObjectId('665453924260c60bfe3e836e'),
ObjectId('665453924260c60bfe3e836f'),
ObjectId('665453924260c60bfe3e8370'),
ObjectId('665453924260c60bfe3e8371'),
ObjectId('665453924260c60bfe3e8372'),
ObjectId('665453924260c60bfe3e8373'),
ObjectId('665453924260c60bfe3e8374'),
ObjectId('665453924260c60bfe3e8375'),
ObjectId('665453924260c60bfe3e8376'),
ObjectId('665453924260c60bfe3e8377'),
ObjectId('665453924260c60bfe3e8378'),
ObjectId('665453924260c60bfe3e8379'),
ObjectId('665453924260c60bfe3e837a'),
ObjectId('665453924260c60bfe3e837b'),
ObjectId('665453924260c60bfe3e837c'),
ObjectId('665453924260c60bfe3e837d'),
ObjectId('665453924260c60bfe3e837e'),
ObjectId('665453924260c60bfe3e837f'),
ObjectId('665453924260c60bfe3e8380'),
ObjectId('665453924260c60bfe3e8381'),
ObjectId('665453924260c60bfe3e8382'),
ObjectId('665453924260c60bfe3e8383'),
ObjectId('665453924260c60bfe3e8384'),
ObjectId('665453924260c60bfe3e8385'),
ObjectId('665453924260c60bfe3e8386'),
ObjectId('665453924260c60bfe3e8387'),
ObjectId('665453924260c60bfe3e8388'),
ObjectId('665453924260c60bfe3e8389'),
ObjectId('665453924260c60bfe3e838a'),
ObjectId('665453924260c60bfe3e838b'),
ObjectId('665453924260c60bfe3e838c'),
ObjectId('665453924260c60bfe3e838d'),
ObjectId('665453924260c60bfe3e838e'),
ObjectId('665453924260c60bfe3e838f'),
ObjectId('665453924260c60bfe3e8390'),
ObjectId('665453924260c60bfe3e8391'),
ObjectId('665453924260c60bfe3e8392'),
ObjectId('665453924260c60bfe3e8393'),
ObjectId('665453924260c60bfe3e8394'),
ObjectId('665453924260c60bfe3e8395'),
ObjectId('665453924260c60bfe3e8396'),
ObjectId('665453924260c60bfe3e8397'),
ObjectId('665453924260c60bfe3e8398'),
ObjectId('665453924260c60bfe3e8399'),
ObjectId('665453924260c60bfe3e839a'),
ObjectId('665453924260c60bfe3e839b'),
ObjectId('665453924260c60bfe3e839c'),
ObjectId('665453924260c60bfe3e839d'),
ObjectId('665453924260c60bfe3e839e'),
ObjectId('665453924260c60bfe3e839f'),
ObjectId('665453924260c60bfe3e83a0'),
ObjectId('665453924260c60bfe3e83a1'),
ObjectId('665453924260c60bfe3e83a2'),
ObjectId('665453924260c60bfe3e83a3'),
ObjectId('665453924260c60bfe3e83a4'),
ObjectId('665453924260c60bfe3e83a5'),
ObjectId('665453924260c60bfe3e83a6'),
ObjectId('665453924260c60bfe3e83a7'),
ObjectId('665453924260c60bfe3e83a8'),
ObjectId('665453924260c60bfe3e83a9'),
ObjectId('665453924260c60bfe3e83aa'),
ObjectId('665453924260c60bfe3e83ab'),
ObjectId('665453924260c60bfe3e83ac'),
ObjectId('665453924260c60bfe3e83ad'),
ObjectId('665453924260c60bfe3e83ae'),
ObjectId('665453924260c60bfe3e83af'),
ObjectId('665453924260c60bfe3e83b0'),
ObjectId('665453924260c60bfe3e83b1'),
ObjectId('665453924260c60bfe3e83b2'),
ObjectId('665453924260c60bfe3e83b3'),
ObjectId('665453924260c60bfe3e83b4'),
ObjectId('665453924260c60bfe3e83b5'),
ObjectId('665453924260c60bfe3e83b6'),
ObjectId('665453924260c60bfe3e83b7'),
ObjectId('665453924260c60bfe3e83b8'),
ObjectId('665453924260c60bfe3e83b9'),
ObjectId('665453924260c60bfe3e83ba'),
ObjectId('665453924260c60bfe3e83bb'),
ObjectId('665453924260c60bfe3e83bc'),
ObjectId('665453924260c60bfe3e83bd'),
ObjectId('665453924260c60bfe3e83be'),
ObjectId('665453924260c60bfe3e83bf'),
ObjectId('665453924260c60bfe3e83c0'),
ObjectId('665453924260c60bfe3e83c1'),
ObjectId('665453924260c60bfe3e83c2'),
ObjectId('665453924260c60bfe3e83c3'),
ObjectId('665453924260c60bfe3e83c4'),
ObjectId('665453924260c60bfe3e83c5'),
ObjectId('665453924260c60bfe3e83c6'),
ObjectId('665453924260c60bfe3e83c7'),
ObjectId('665453924260c60bfe3e83c8'),
ObjectId('665453924260c60bfe3e83c9'),
ObjectId('665453924260c60bfe3e83ca'),
ObjectId('665453924260c60bfe3e83cb'),
ObjectId('665453924260c60bfe3e83cc'),
ObjectId('665453924260c60bfe3e83cd'),
ObjectId('665453924260c60bfe3e83ce'),
ObjectId('665453924260c60bfe3e83cf'),
ObjectId('665453924260c60bfe3e83d0'),
ObjectId('665453924260c60bfe3e83d1'),
ObjectId('665453924260c60bfe3e83d2'),
ObjectId('665453924260c60bfe3e83d3'),
ObjectId('665453924260c60bfe3e83d4'),
ObjectId('665453924260c60bfe3e83d5'),
ObjectId('665453924260c60bfe3e83d6'),
ObjectId('665453924260c60bfe3e83d7'),
ObjectId('665453924260c60bfe3e83d8'),
ObjectId('665453924260c60bfe3e83d9'),
ObjectId('665453924260c60bfe3e83da'),
ObjectId('665453924260c60bfe3e83db'),
ObjectId('665453924260c60bfe3e83dc'),
ObjectId('665453924260c60bfe3e83dd'),
ObjectId('665453924260c60bfe3e83de'),
ObjectId('665453924260c60bfe3e83df'),
ObjectId('665453924260c60bfe3e83e0'),
ObjectId('665453924260c60bfe3e83e1'),
ObjectId('665453924260c60bfe3e83e2'),
ObjectId('665453924260c60bfe3e83e3'),
ObjectId('665453924260c60bfe3e83e4'),
ObjectId('665453924260c60bfe3e83e5'),
ObjectId('665453924260c60bfe3e83e6'),
ObjectId('665453924260c60bfe3e83e7'),
ObjectId('665453924260c60bfe3e83e8'),
ObjectId('665453924260c60bfe3e83e9'),
ObjectId('665453924260c60bfe3e83ea'),
ObjectId('665453924260c60bfe3e83eb'),
ObjectId('665453924260c60bfe3e83ec'),
ObjectId('665453924260c60bfe3e83ed'),
ObjectId('665453924260c60bfe3e83ee'),
ObjectId('665453924260c60bfe3e83ef'),
ObjectId('665453924260c60bfe3e83f0'),
ObjectId('665453924260c60bfe3e83f1'),
ObjectId('665453924260c60bfe3e83f2'),
ObjectId('665453924260c60bfe3e83f3'),
ObjectId('665453924260c60bfe3e83f4'),
ObjectId('665453924260c60bfe3e83f5'),
ObjectId('665453924260c60bfe3e83f6'),
ObjectId('665453924260c60bfe3e83f7'),
ObjectId('665453924260c60bfe3e83f8'),
ObjectId('665453924260c60bfe3e83f9'),
ObjectId('665453924260c60bfe3e83fa'),
ObjectId('665453924260c60bfe3e83fb'),
ObjectId('665453924260c60bfe3e83fc'),
ObjectId('665453924260c60bfe3e83fd'),
ObjectId('665453924260c60bfe3e83fe'),
ObjectId('665453924260c60bfe3e83ff'),
ObjectId('665453924260c60bfe3e8400'),
ObjectId('665453924260c60bfe3e8401'),
ObjectId('665453924260c60bfe3e8402'),
ObjectId('665453924260c60bfe3e8403'),
ObjectId('665453924260c60bfe3e8404'),
ObjectId('665453924260c60bfe3e8405'),
ObjectId('665453924260c60bfe3e8406'),
ObjectId('665453924260c60bfe3e8407'),
ObjectId('665453924260c60bfe3e8408'),
ObjectId('665453924260c60bfe3e8409'),
ObjectId('665453924260c60bfe3e840a'),
ObjectId('665453924260c60bfe3e840b'),
ObjectId('665453924260c60bfe3e840c'),
ObjectId('665453924260c60bfe3e840d'),
ObjectId('665453924260c60bfe3e840e'),
ObjectId('665453924260c60bfe3e840f'),
ObjectId('665453924260c60bfe3e8410'),
ObjectId('665453924260c60bfe3e8411'),
ObjectId('665453924260c60bfe3e8412'),
ObjectId('665453924260c60bfe3e8413'),
ObjectId('665453924260c60bfe3e8414'),
ObjectId('665453924260c60bfe3e8415'),
ObjectId('665453924260c60bfe3e8416'),
ObjectId('665453924260c60bfe3e8417'),
ObjectId('665453924260c60bfe3e8418'),
ObjectId('665453924260c60bfe3e8419'),
ObjectId('665453924260c60bfe3e841a'),
ObjectId('665453924260c60bfe3e841b'),
ObjectId('665453924260c60bfe3e841c'),
ObjectId('665453924260c60bfe3e841d'),
ObjectId('665453924260c60bfe3e841e'),
ObjectId('665453924260c60bfe3e841f'),
ObjectId('665453924260c60bfe3e8420'),
ObjectId('665453924260c60bfe3e8421'),
ObjectId('665453924260c60bfe3e8422'),
ObjectId('665453924260c60bfe3e8423'),
ObjectId('665453924260c60bfe3e8424'),
ObjectId('665453924260c60bfe3e8425'),
ObjectId('665453924260c60bfe3e8426'),
ObjectId('665453924260c60bfe3e8427'),
ObjectId('665453924260c60bfe3e8428'),
ObjectId('665453924260c60bfe3e8429'),
ObjectId('665453924260c60bfe3e842a'),
ObjectId('665453924260c60bfe3e842b'),
ObjectId('665453924260c60bfe3e842c'),
ObjectId('665453924260c60bfe3e842d'),
ObjectId('665453924260c60bfe3e842e'),
ObjectId('665453924260c60bfe3e842f'),
ObjectId('665453924260c60bfe3e8430'),
ObjectId('665453924260c60bfe3e8431'),
ObjectId('665453924260c60bfe3e8432'),
ObjectId('665453924260c60bfe3e8433'),
ObjectId('665453924260c60bfe3e8434'),
ObjectId('665453924260c60bfe3e8435'),
ObjectId('665453924260c60bfe3e8436'),
ObjectId('665453924260c60bfe3e8437'),
ObjectId('665453924260c60bfe3e8438'),
ObjectId('665453924260c60bfe3e8439'),
ObjectId('665453924260c60bfe3e843a'),
ObjectId('665453924260c60bfe3e843b'),
ObjectId('665453924260c60bfe3e843c'),
ObjectId('665453924260c60bfe3e843d'),
ObjectId('665453924260c60bfe3e843e'),
ObjectId('665453924260c60bfe3e843f'),
ObjectId('665453924260c60bfe3e8440'),
ObjectId('665453924260c60bfe3e8441'),
ObjectId('665453924260c60bfe3e8442'),
ObjectId('665453924260c60bfe3e8443'),
ObjectId('665453924260c60bfe3e8444'),
ObjectId('665453924260c60bfe3e8445'),
ObjectId('665453924260c60bfe3e8446'),
ObjectId('665453924260c60bfe3e8447'),
ObjectId('665453924260c60bfe3e8448'),
ObjectId('665453924260c60bfe3e8449'),
ObjectId('665453924260c60bfe3e844a'),
ObjectId('665453924260c60bfe3e844b'),
ObjectId('665453924260c60bfe3e844c'),
ObjectId('665453924260c60bfe3e844d'),
ObjectId('665453924260c60bfe3e844e'),
ObjectId('665453924260c60bfe3e844f'),
ObjectId('665453924260c60bfe3e8450'),
ObjectId('665453924260c60bfe3e8451'),
ObjectId('665453924260c60bfe3e8452'),
ObjectId('665453924260c60bfe3e8453'),
ObjectId('665453924260c60bfe3e8454'),
ObjectId('665453924260c60bfe3e8455'),
ObjectId('665453924260c60bfe3e8456'),
ObjectId('665453924260c60bfe3e8457'),
ObjectId('665453924260c60bfe3e8458'),
ObjectId('665453924260c60bfe3e8459'),
ObjectId('665453924260c60bfe3e845a'),
ObjectId('665453924260c60bfe3e845b'),
ObjectId('665453924260c60bfe3e845c'),
ObjectId('665453924260c60bfe3e845d'),
ObjectId('665453924260c60bfe3e845e'),
ObjectId('665453924260c60bfe3e845f'),
ObjectId('665453924260c60bfe3e8460'),
ObjectId('665453924260c60bfe3e8461'),
ObjectId('665453924260c60bfe3e8462'),
ObjectId('665453924260c60bfe3e8463'),
ObjectId('665453924260c60bfe3e8464'),
ObjectId('665453924260c60bfe3e8465'),
ObjectId('665453924260c60bfe3e8466'),
ObjectId('665453924260c60bfe3e8467'),
ObjectId('665453924260c60bfe3e8468'),
ObjectId('665453924260c60bfe3e8469'),
ObjectId('665453924260c60bfe3e846a')],
TaskWorkflow(database=<superduperdb.base.datalayer.Datalayer object at 0x13ed87e90>, G=<networkx.classes.digraph.DiGraph object at 0x13db43190>))

We can verify that the images are correctly saved by retrieved a single record:

r = db['images'].find_one().execute()
r

Outputs

Document({'img': <PIL.PngImagePlugin.PngImageFile image mode=RGB size=500x338 at 0x13EE96590>, '_fold': 'train', '_schema': 'AUTO:img=pil_image', '_id': ObjectId('665453924260c60bfe3e8084')})

The contents of the Document may be accessed by calling .unpack(). You can see that the images were saved and retrieved correctly.

r.unpack()['img']

Outputs

We now build a torch model for text-2-image similarity using the clip library. In order to save the outputs correctly in the system, we add the tensor datatype to the model:

import clip
import torch
from superduperdb.ext.torch import TorchModel, tensor


model, preprocess = clip.load("ViT-B/32", "cpu")

class ImageModel(torch.nn.Module):
    def __init__(self):
        super().__init__()
        self.model = model

    def forward(self, image_tensors):
        return self.model.encode_image(image_tensors)


dt = tensor(dtype='float', shape=(512,))


image_model = TorchModel(
    identifier='clip_image',
    object=ImageModel(),
    preprocess=preprocess,
    datatype=dt,
    loader_kwargs={'batch_size': 5},
)

Outputs

We can verify that this model gives us the correct outputs on the added data with the .predict method:

image_model.predict(data[0]['img'])

Outputs

tensor([ 1.1590e-01, -1.1163e-01, 1.6210e-01, 3.2125e-01, 3.8310e-02, -2.9220e-01, -1.6120e-01, -1.0013e+00, 4.9538e-01, 1.4114e-01, 7.6920e-02, 6.7780e-02, -1.0894e-01, 1.2793e-01, -2.6868e-01, 4.6839e-01, 5.3715e-01, 7.9151e-02, 2.9155e-02, 2.5880e-01, -3.8380e-01, 8.6311e-02, 1.8946e-01, 1.6239e-01, -6.7896e-01, 6.5299e-02, 4.9489e-01, 1.5839e-01, 8.3728e-02, 5.5632e-02, -1.0379e-01, 1.6675e-02, -3.3331e-01, 1.2236e-01, 6.2966e-01, -2.9543e-01, 6.5257e-01, -6.9910e-02, 2.0470e-01, 1.9606e+00, -1.2133e-01, -5.8945e-02, -1.3498e-01, -1.3249e-01, 2.0738e-01, -7.0674e-01, 1.3906e-01, 1.7988e-01, 7.0238e-02, -3.3584e-01, -2.3665e-01, 7.9334e-02, -1.0090e-01, -9.1650e-02, 1.7352e-01, 1.1216e-01, 1.9300e-01, 4.8928e-01, -1.1548e-01, 5.7670e-02, 6.2232e-01, -3.1829e-01, -2.3148e-01, 2.0030e-01, -4.0209e-02, -1.4554e-01, 4.4466e-01, 4.1464e-01, -2.4179e-01, -1.2451e-01, 4.8724e-01, -4.7054e-02, 2.0541e-02, -4.9354e-03, 8.7666e-02, 1.9785e-02, -1.6447e-01, -8.3285e-01, -1.2699e-01, 1.2998e-01, -2.3226e-01, -8.8556e-01, -1.7331e-01, -7.3583e-01, 1.2145e-01, -9.4184e-02, 9.7923e-01, -4.3598e-01, 3.9887e-01, 1.7900e-01, 4.4396e-02, -4.5832e-01, -6.0410e+00, -4.3534e-01, -5.9481e-01, 4.9649e-02, -2.7357e-02, -3.8371e-01, -9.3306e-01, 9.2320e-01, -1.9619e-01, -3.2803e-02, -1.9339e-02, -8.1225e-02, 4.7819e-01, 3.1278e-02, 1.9959e-01, -2.0033e-01, -1.1671e-01, -3.9358e-01, -1.2317e-01, -9.5839e-01, 5.1610e-01, 1.2667e-01, 5.3419e-02, -8.2035e-02, 2.4216e-01, 7.0817e-03, 1.1345e-01, 5.7588e-02, 3.5402e-01, -4.0607e-01, -4.1502e-02, -4.2697e-01, 3.4891e-01, -5.3146e-01, -1.0416e-01, 2.7876e-01, 1.4609e-01, -9.3855e-02, -1.9221e-01, -1.3139e-01, -1.7017e-01, 7.5767e-01, -4.1235e-01, 1.9602e-01, -4.5141e-01, -2.7854e-01, 3.4447e-01, 2.3144e-02, 5.0284e-01, 7.0880e-02, -4.6520e-02, 4.0040e-01, 1.0449e-01, 5.3388e-01, 4.8889e-01, -1.3331e-01, 3.1072e-01, 1.7776e-01, -1.4810e-01, 3.3747e-01, -5.3392e-01, -1.0897e-01, -1.0275e-01, -1.0211e-01, -2.4024e-01, -9.3492e-02, 1.4864e-01, -5.1577e-01, 5.0001e-01, 3.8927e-01, -2.4130e-01, -2.1834e-01, 2.5411e-01, -3.3644e-02, -2.9853e-01, 2.5758e-01, -1.6819e-01, 3.2913e-02, -2.6224e-01, 7.2353e-02, -1.1571e-01, 1.2307e-01, 5.3886e-02, -3.0080e-01, 7.7794e-01, 3.1006e-01, 1.4186e-01, 3.2553e-01, 1.8123e-01, 5.2372e-02, -3.1844e-01, 9.4648e-03, 2.7061e-01, 1.3351e-02, 1.5009e-01, 3.6278e-02, 3.6682e-01, 3.1048e-02, 1.9435e-01, -5.7764e-02, 7.7549e-01, -1.4513e-01, 2.3016e-01, -3.9774e-02, -4.3843e-01, -2.7769e-01, -1.1522e+00, -7.6351e-02, -1.1522e-01, -1.1125e-01, 1.3012e-01, 3.2185e-01, -4.1987e-01, 1.1690e-01, 2.2993e-01, -5.4273e-02, 7.1026e-02, 6.8535e-02, -3.1000e-01, 7.4613e-01, -1.7190e-01, 6.3011e-01, -1.7141e-02, -3.8054e-01, -2.2758e-01, -1.2865e-01, 9.5331e-01, -1.4831e-01, 4.1659e-01, 3.2765e-01, -1.7633e-01, -4.9092e-02, -5.7158e-01, 1.1250e-01, 8.9816e-02, -1.1677e-01, 3.0087e-01, -2.5121e-01, 2.1989e-01, -4.8331e-01, -5.2032e-01, 8.8859e-02, 1.5607e-01, 4.7345e-01, -2.8604e-01, 6.8347e-01, 3.0517e-02, -2.5008e-01, 2.9491e-01, -2.0136e-01, -1.6408e-01, 7.5016e-02, 1.4922e-01, 7.9002e-01, 3.8315e-02, -7.2039e-01, -2.8794e-01, 2.5925e-01, -4.6311e-01, -8.0276e-02, -2.5208e-01, -1.8260e-01, 1.0297e-01, -4.0524e-02, -4.5251e-04, -8.3430e-02, 4.3200e-01, 3.4515e-01, -5.0279e-01, 9.2067e-02, -2.5708e-01, -1.5032e+00, -3.8652e-02, -4.5196e-01, 8.5718e-03, -2.4990e-01, -3.9936e-02, 1.2828e-02, -3.1357e-01, 1.5526e-01, 5.4366e-01, 1.0955e-01, -5.4420e-01, -8.8090e-02, -2.4157e-01, -1.6538e-01, -3.1087e-01, 5.2663e-02, 2.5429e-01, -4.8138e-02, -4.8485e-01, 3.8829e-01, -1.6188e-01, -1.2533e-01, -3.6505e-01, -1.3971e-01, -3.6999e-01, -8.9621e-02, 1.0524e-01, 4.7519e-02, 1.4970e-01, -3.3221e-02, 5.3704e-02, 7.2312e-02, 4.6348e-01, -1.5002e-02, 4.2172e-01, 2.0594e-01, 1.4226e-01, 3.5545e-01, 2.2918e-02, -2.2476e-01, -3.0733e-03, -3.2779e-01, -5.9857e-02, 2.0830e-01, 5.2765e-01, 5.2056e-02, -3.0684e-01, 1.7120e-01, 7.5580e-01, 2.8488e-01, -1.0147e-01, 2.9288e-01, 2.9232e-01, 5.2558e-01, 2.6853e-01, -6.0759e-01, 3.1820e-01, 8.0559e-01, -2.9796e-01, -2.1610e-01, 3.1361e-02, 3.7303e-02, -6.2717e-02, 5.5843e-02, 8.4793e-02, -2.6605e-01, 9.1591e-02, -3.2060e-01, -3.1954e-01, -2.4636e-01, 4.0276e-01, 4.6442e-01, -3.3635e-01, -1.8236e-01, -3.7923e-02, 2.3812e-01, 2.5474e-02, -3.7982e-01, -3.2390e-01, 4.8698e-01, 4.0739e-01, -2.6965e-01, 3.4426e-02, 7.5702e-02, 3.2427e-01, -7.2199e-02, -2.7115e-01, -2.5527e-01, 5.9944e-01, -1.5569e-01, -9.0987e-02, 4.4571e-01, -1.8296e-01, -4.9801e-01, 1.4001e-01, 6.0739e-01, -2.1250e-01, -3.2282e-02, 6.0238e-01, 1.1127e-01, -1.2075e+00, 1.1658e-01, -1.8768e-01, -1.4897e-01, 4.3965e-01, -4.7514e-02, -1.1361e-01, 3.4682e-01, 7.2601e-02, 2.5298e-01, -5.5959e-01, -2.2657e-01, 1.0485e+00, 5.7782e-01, -6.8249e-01, 2.0709e-01, 3.0597e-01, -4.3153e-01, -2.5387e-01, 3.8834e-01, 5.1551e-01, 5.2100e-03, 3.3520e-01, -6.7791e-01, 6.6559e-01, -1.2692e+00, -1.4179e-01, -2.5137e-01, 2.1868e-01, -4.4603e-01, 8.2985e-02, 3.0630e-01, 1.5204e-01, -1.2356e-01, -4.1893e-01, -6.5163e-01, -1.3898e-01, -3.1360e-01, 3.4516e-01, 4.4857e-01, -3.9492e-01, 7.7293e-02, -4.2672e-02, -6.2462e-01, 7.2953e-01, -2.3767e-01, 5.9171e-01, -1.5064e-01, -3.6149e-01, 2.1571e-01, -6.6212e-02, -3.1349e-01, 2.2127e-01, -1.1634e-01, 5.9028e-02, -5.0800e-01, -2.1432e-01, 1.0446e+00, 8.5178e-02, 2.9323e-01, 2.8574e-01, 3.2103e-01, -2.4577e-01, 2.7432e-01, 1.1078e+00, -8.9108e-02, 1.6229e-01, 1.6318e-01, 1.3020e-01, -1.5229e-02, 2.8989e-01, -5.8030e-01, -8.8477e-02, 4.3783e-01, 2.3567e-01, 1.1670e-01, -2.0998e-01, 2.5156e-01, -2.3052e-01, -2.1780e-01, 6.5135e-01, -1.6850e-01, 7.4846e-02, -1.8685e-01, 1.2253e-01, -4.1852e-01, -5.9780e-02, -7.0781e-01, -6.5484e-01, 7.1986e-01, -1.1047e-01, 1.1983e-01, 5.5121e-01, 3.3051e-01, 4.2507e-01, 2.3734e-01, 8.2319e-02, -2.0200e-01, 5.3302e-02, 5.6774e-01, 8.5539e-02, 3.8295e-01, 1.8800e-02, -9.5386e-01, 7.4490e-02, 2.6489e-01, 5.4485e-01, -1.8950e-01, -3.4931e-01, -8.4142e-02, -5.1462e-01, -5.5838e-02, 3.9012e-02, 2.8505e-01, 4.5606e-01, -2.2018e-01, 1.4774e-01, 8.0736e-02, -6.9549e-02, 3.5784e-01, -2.9284e-01, -6.1763e-02, -5.6776e-02, -3.8342e-01, -3.8281e-01, 1.0981e-01, 3.9572e-01, -3.6426e-01, 1.0481e-01, -2.7062e-01, 1.5884e-01, 3.8384e-02, 1.0798e-01, 3.0277e-01, 2.1358e-01, 6.2604e-02, -3.6589e-01, 6.0148e-02, -2.0930e-01, 3.1359e-01, 7.2983e-01, -1.5445e-01, -5.2086e-02, -1.5287e-01, 3.1047e-01, 1.3831e+00, 3.4753e-01, 2.2045e-01])

Now we'd like to set up this model to compute outputs on the 'img' key of each record. To do that we create a Listener (see here for more information) which "listens" for incoming and existing data, and computes outputs on that data.

When new data is inserted, the model automatically will create outputs on that data. This is a very handy feature for productionizing AI and ML, since a data deployment needs to be keep up-to-date as far as possible.

listener = image_model.to_listener(
    select=db['images'].find(),
    key='img',
    identifier='image_predictions',
)

_ = db.apply(listener)

Outputs

2024-May-27 11:34:16.44| WARNING  | Duncans-MBP.fritz.box| superduperdb.backends.local.artifacts:82   | File /tmp/e1635b227a7f3787dc79524d812915c342701260 already exists
2024-May-27 11:34:17.37| INFO     | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:37   | Submitting job. function:<function method_job at 0x10cd4bec0>

999it [00:00, 107213.29it/s] 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 200/200 [00:30<00:00, 6.58it/s]

2024-May-27 11:34:49.03| INFO     | Duncans-MBP.fritz.box| superduperdb.components.model:783  | Adding 999 model outputs to db
2024-May-27 11:34:49.67| WARNING  | Duncans-MBP.fritz.box| superduperdb.backends.mongodb.query:254  | Some delete ids are not executed , hence halting execution Please note the partially executed operations wont trigger any model/listeners unless CDC is active.
2024-May-27 11:34:49.76| SUCCESS  | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:43   | Job submitted on <superduperdb.backends.local.compute.LocalComputeBackend object at 0x13ecc76d0>.  function:<function method_job at 0x10cd4bec0> future:5d4a6013-900c-4582-9680-4043e1407519

We can verify that the outputs are correctly inserted into the documents with this query. The outputs are saved in the listener.outputs field:

list(listener.outputs_select.limit(1).execute())[0][listener.outputs].unpack()

Outputs

/Users/dodo/SuperDuperDB/superduperdb/superduperdb/ext/torch/encoder.py:52: UserWarning: The given NumPy array is not writable, and PyTorch does not support non-writable tensors. This means writing to this tensor will result in undefined behavior. You may want to copy the array to protect its data or make it writable before converting it to a tensor. This type of warning will be suppressed for the rest of this program. (Triggered internally at /Users/runner/work/pytorch/pytorch/pytorch/torch/csrc/utils/tensor_numpy.cpp:212.)
return torch.from_numpy(array)

Downstream of this first model, we now can add another smaller model, to classify images with configurable terms. Since the dataset is concerned with cats and dogs we create 2 downstream models classifying the images in 2 different ways.

from superduperdb import ObjectModel


class Comparer:
    def __init__(self, words, text_features):
        self.targets = {w: text_features[i] for i, w in enumerate(words)}
        self.lookup = list(self.targets.keys())
        self.matrix = torch.stack(list(self.targets.values()))

    def __call__(self, vector):
        best = (self.matrix @ vector).topk(1)[1].item()
        return self.lookup[best]


cats_vs_dogs = ObjectModel(
    'cats_vs_dogs',
    object=Comparer(['cat', 'dog'], model.encode_text(clip.tokenize(['cat', 'dog']))),
).to_listener(
    select=db['images'].find(),
    key=listener.outputs,
)

            
felines_vs_canines = ObjectModel(
    'felines_vs_canines',
    object=Comparer(['feline', 'canine'], model.encode_text(clip.tokenize(['feline', 'canine']))),
).to_listener(
    select=db['images'].find(),
    key=listener.outputs,
)


db.apply(cats_vs_dogs)
db.apply(felines_vs_canines)

Outputs

2024-May-27 11:34:50.07| INFO     | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:37   | Submitting job. function:<function method_job at 0x10cd4bec0>

999it [00:00, 130533.01it/s]

2024-May-27 11:34:50.63| INFO     | Duncans-MBP.fritz.box| superduperdb.components.model:783  | Adding 999 model outputs to db
2024-May-27 11:34:51.22| WARNING  | Duncans-MBP.fritz.box| superduperdb.backends.mongodb.query:254  | Some delete ids are not executed , hence halting execution Please note the partially executed operations wont trigger any model/listeners unless CDC is active.
2024-May-27 11:34:51.22| SUCCESS  | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:43   | Job submitted on <superduperdb.backends.local.compute.LocalComputeBackend object at 0x13ecc76d0>.  function:<function method_job at 0x10cd4bec0> future:3c5da5ea-a4d2-4964-89ed-5691e0e80d4b
2024-May-27 11:34:51.23| INFO     | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:37   | Submitting job. function:<function method_job at 0x10cd4bec0>

999it [00:00, 125897.17it/s]

2024-May-27 11:34:51.51| INFO     | Duncans-MBP.fritz.box| superduperdb.components.model:783  | Adding 999 model outputs to db
2024-May-27 11:34:52.08| WARNING  | Duncans-MBP.fritz.box| superduperdb.backends.mongodb.query:254  | Some delete ids are not executed , hence halting execution Please note the partially executed operations wont trigger any model/listeners unless CDC is active.
2024-May-27 11:34:52.08| SUCCESS  | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:43   | Job submitted on <superduperdb.backends.local.compute.LocalComputeBackend object at 0x13ecc76d0>.  function:<function method_job at 0x10cd4bec0> future:f46f6471-7895-482d-ab03-ec7344235512

([<superduperdb.jobs.job.ComponentJob at 0x142c03d90>],
Listener(identifier='component/listener/felines_vs_canines/e92d248d-8ea8-4e2b-8254-5988acfea072', uuid='e92d248d-8ea8-4e2b-8254-5988acfea072', key='_outputs.b55cc5c5-9427-40ee-95a3-227e659cd783', model=ObjectModel(identifier='felines_vs_canines', uuid='65f2761b-9d4c-47fb-a5a9-f5e95c81bda9', signature='*args,**kwargs', datatype=None, output_schema=None, flatten=False, model_update_kwargs={}, predict_kwargs={}, compute_kwargs={}, validation=None, metric_values={}, num_workers=0, object=<main.Comparer object at 0x177f42bd0>), select=images.find(), active=True, predict_kwargs={}))

We can verify that both downstream models have written their outputs to the database by querying a document:

r = db['images'].find_one().execute()

print(r[cats_vs_dogs.outputs])
print(r[felines_vs_canines.outputs])

Outputs

cat feline

Let's check that the predictions make sense for the inserted images:

db['images'].find_one({cats_vs_dogs.outputs: 'cat'}).execute()['img']

Outputs

db['images'].find_one({felines_vs_canines.outputs: 'feline'}).execute()['img']

Outputs

db['images'].find_one({cats_vs_dogs.outputs: 'dog'}).execute()['img']

Outputs

db['images'].find_one({felines_vs_canines.outputs: 'canine'}).execute()['img']

Outputs

Now that we have installed the models using Listener, we can insert new data. This data should be automatically processed by the installed models:

db['images'].insert_one(Document({**data[-1], 'new': True})).execute()

Outputs

2024-May-27 11:36:59.83| WARNING  | Duncans-MBP.fritz.box| superduperdb.misc.annotations:117  | add is deprecated and will be removed in a future release.
2024-May-27 11:37:00.62| INFO     | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:37   | Submitting job. function:<function callable_job at 0x10cd4be20>
2024-May-27 11:37:00.62| SUCCESS  | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:43   | Job submitted on <superduperdb.backends.local.compute.LocalComputeBackend object at 0x13ecc76d0>.  function:<function callable_job at 0x10cd4be20> future:ce572425-a6bd-4170-bad9-6ff56598388d
2024-May-27 11:37:00.90| INFO     | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:37   | Submitting job. function:<function method_job at 0x10cd4bec0>

1it [00:00, 305.06it/s] 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 11.75it/s]

2024-May-27 11:37:01.54| INFO     | Duncans-MBP.fritz.box| superduperdb.components.model:783  | Adding 1 model outputs to db
2024-May-27 11:37:01.54| WARNING  | Duncans-MBP.fritz.box| superduperdb.backends.mongodb.query:254  | Some delete ids are not executed , hence halting execution Please note the partially executed operations wont trigger any model/listeners unless CDC is active.
2024-May-27 11:37:01.54| SUCCESS  | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:43   | Job submitted on <superduperdb.backends.local.compute.LocalComputeBackend object at 0x13ecc76d0>.  function:<function method_job at 0x10cd4bec0> future:189c7a61-61f3-4fbc-837c-4bdd38d6241d
2024-May-27 11:37:01.54| INFO     | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:37   | Submitting job. function:<function method_job at 0x10cd4bec0>

1it [00:00, 319.59it/s]

2024-May-27 11:37:01.55| INFO     | Duncans-MBP.fritz.box| superduperdb.components.model:783  | Adding 1 model outputs to db
2024-May-27 11:37:01.55| WARNING  | Duncans-MBP.fritz.box| superduperdb.backends.mongodb.query:254  | Some delete ids are not executed , hence halting execution Please note the partially executed operations wont trigger any model/listeners unless CDC is active.
2024-May-27 11:37:01.55| SUCCESS  | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:43   | Job submitted on <superduperdb.backends.local.compute.LocalComputeBackend object at 0x13ecc76d0>.  function:<function method_job at 0x10cd4bec0> future:1c603b81-4891-4d9b-8c1b-48039e77e27d
2024-May-27 11:37:01.55| INFO     | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:37   | Submitting job. function:<function method_job at 0x10cd4bec0>

1it [00:00, 322.39it/s]

2024-May-27 11:37:01.56| INFO     | Duncans-MBP.fritz.box| superduperdb.components.model:783  | Adding 1 model outputs to db
2024-May-27 11:37:01.56| WARNING  | Duncans-MBP.fritz.box| superduperdb.backends.mongodb.query:254  | Some delete ids are not executed , hence halting execution Please note the partially executed operations wont trigger any model/listeners unless CDC is active.
2024-May-27 11:37:01.56| SUCCESS  | Duncans-MBP.fritz.box| superduperdb.backends.local.compute:43   | Job submitted on <superduperdb.backends.local.compute.LocalComputeBackend object at 0x13ecc76d0>.  function:<function method_job at 0x10cd4bec0> future:a5b04a66-feac-4270-8685-edb0de90142f

([ObjectId('6654543b4260c60bfe3e8479')],
TaskWorkflow(database=<superduperdb.base.datalayer.Datalayer object at 0x13ed87e90>, G=<networkx.classes.digraph.DiGraph object at 0x15c991d10>))

We can verify this by querying the data again:

r = db['images'].find_one({'new': True}).execute().unpack()
r['img']

Outputs

You see here that the models have been called in the correct order on the newly added data and the outputs saved to the new record:

r['_outputs']

Outputs

{'b55cc5c5-9427-40ee-95a3-227e659cd783': tensor([-1.9784e-01, -1.1985e-02, 1.5070e-02, 2.3934e-01, 9.1826e-02, -6.8618e-01, 5.9504e-01, -1.3494e+00, -1.2732e-01, 1.8213e-01, -2.4697e-01, 4.2079e-01, 5.8837e-03, -6.9505e-02, 3.9881e-01, -1.6329e-01, 1.0990e+00, -2.4527e-01, 4.1645e-01, 2.2653e-01, -4.2598e-01, -1.3891e-01, 3.8267e-01, -4.2262e-01, 4.2060e-02, 1.8807e-01, 5.9752e-01, 2.3998e-01, 5.2914e-01, 6.6496e-02, 2.9223e-01, 3.7576e-01, -2.6834e-01, -3.8394e-01, -5.1581e-01, 6.0321e-02, 4.3297e-01, -9.6605e-02, -4.9369e-01, 1.7737e+00, -5.3931e-01, -3.9127e-01, 1.7854e-01, -2.3078e-01, 2.8036e-01, 1.3256e-01, 1.7078e-01, -1.3245e-02, -1.1761e-01, 4.4366e-02, -2.0976e-01, 4.0344e-02, -3.3105e-02, 1.1045e-01, 4.9561e-01, 1.8460e-02, -4.0324e-01, 2.0999e-01, -8.3989e-02, -4.0953e-01, 6.7217e-01, 2.4515e-01, 7.3640e-01, -5.4399e-02, -3.6352e-01, 3.7475e-02, 2.2933e-01, 2.1635e-01, -3.7907e-01, 7.3136e-02, 1.7207e-01, -2.2663e-01, 2.0013e-01, 3.5541e-01, -3.6459e-01, -1.5265e-01, -2.8436e-01, -3.8413e-01, -2.1256e-02, 1.6117e-01, 8.1823e-02, -2.8004e-01, 5.4281e-02, -6.5066e-01, 4.6435e-01, -3.6090e-01, 1.4236e+00, -3.5425e-01, -2.0861e-01, 1.5555e-01, -2.3906e-01, -2.6277e-01, -6.1501e+00, 6.8107e-01, -4.6798e-01, -2.5333e-01, 1.2230e-01, 8.8723e-02, -1.0091e+00, 1.0498e+00, -1.6454e-01, -1.4257e-01, -7.2294e-02, -1.9788e-01, 2.5033e-01, 9.9614e-02, 1.6566e-01, -2.6232e-01, -2.2362e-01, -3.7903e-01, -2.1926e-01, -4.8382e-01, 4.1420e-01, -2.5579e-02, 4.6423e-02, 3.9816e-01, -2.2495e-01, -2.0562e-01, -4.5065e-01, 1.1380e-01, 5.0083e-01, -3.8125e-01, 1.2009e-01, -1.7179e-01, -9.2985e-02, -9.0663e-01, 1.1145e-01, 4.6361e-01, -4.8051e-01, -3.1573e-01, 3.9665e-01, 8.7184e-02, -4.5003e-01, 7.2500e-01, -2.8902e-01, 2.6280e-01, 1.4857e-01, -4.4423e-02, -3.3938e-01, -2.3859e-01, 4.9975e-02, -2.1402e-01, -3.5357e-01, 1.9798e-01, -3.5788e-01, 2.6473e-01, 3.6177e-01, 3.8417e-01, -1.6142e-01, 4.3199e-02, -3.2399e-01, 3.5889e-01, -6.3973e-01, 3.7262e-01, -8.1334e-01, 2.0936e-01, -5.3070e-01, 1.2967e-01, 9.4323e-02, 1.4453e-01, -2.8726e-01, 3.4551e-01, -4.3649e-02, 2.1140e-01, 3.9445e-01, -3.2790e-01, -4.2833e-01, 2.5720e-01, 9.5053e-02, 4.3625e-01, -5.7717e-02, 2.2466e-01, 7.9365e-02, 4.0048e-02, 2.5109e-01, 1.0167e-01, 1.4203e+00, -2.5085e-01, 2.8730e-01, -2.0506e-01, -6.9146e-02, 2.2863e-01, 5.7341e-01, -4.3942e-02, -3.0144e-01, -6.5863e-02, -3.5006e-01, 3.7668e-01, -2.0926e-01, 2.0152e-01, -2.4407e-03, -5.2490e-02, 2.6947e-01, 1.4438e-01, 2.8786e-01, 2.7246e-01, 3.1032e-01, -2.0468e-01, -1.1268e+00, -5.5739e-01, 4.1084e-02, -1.0877e-01, 5.1607e-01, 8.3844e-02, -3.7792e-01, -4.8277e-01, 1.9685e-01, -2.5837e-01, -3.4591e-01, 3.5694e-01, -4.3230e-01, 1.0324e-01, -4.7433e-01, 3.6803e-03, 9.4960e-02, -2.3161e-01, -3.2248e-01, 2.6306e-01, 1.1879e+00, 4.2923e-02, 6.0710e-01, 3.5460e-01, 9.0147e-02, 2.4508e-01, -4.2532e-01, -2.7915e-01, 4.6251e-01, -3.4883e-01, -8.6505e-02, 4.5012e-01, 3.1542e-01, 6.6807e-02, -4.2769e-01, 3.6456e-01, -2.1610e-01, 6.7809e-01, 3.0854e-01, -1.8361e-01, 1.9379e-02, 5.7221e-02, 2.3852e-01, 1.4775e-01, -5.2261e-02, 1.0130e-02, -1.5239e-01, 8.6012e-01, 4.6111e-01, -6.5358e-01, 3.6930e-01, 4.3411e-01, 4.9753e-01, 8.0125e-02, 2.4168e-01, -5.1495e-02, 3.2372e-01, -1.3840e-01, -4.4837e-01, -1.1076e-01, 1.4098e+00, 2.3750e-01, 1.2766e-01, 2.1274e-01, 5.0297e-01, -1.1665e+00, 2.7810e-01, -3.8077e-01, -1.8598e-01, -3.1121e-01, -3.7083e-01, 2.4960e-01, -4.1929e-01, -2.4243e-01, 3.9122e-01, 1.7913e-02, 2.7623e-01, 7.9261e-02, -1.8902e-02, 3.6915e-01, 2.3819e-01, 4.9772e-02, -3.8237e-01, -6.3968e-02, -7.0589e-01, -1.1346e-01, -3.8071e-01, -2.6041e-01, 7.2083e-01, 4.2882e-02, -2.5185e-01, 7.4913e-02, 2.4498e-02, -1.2440e-01, -1.3673e-01, 4.0574e-01, 1.1170e-01, -1.4629e-01, 4.2372e-01, 5.7550e-01, 2.1063e-01, 2.1003e-01, -6.9390e-01, 1.2443e-01, -1.6181e-01, 4.6245e-02, 1.4564e-01, -1.9245e-01, 1.8174e-01, -8.2501e-02, 7.1926e-03, 4.4878e-01, -2.2032e-01, 6.9630e-01, 7.2316e-01, -3.4519e-01, 9.0801e-02, 1.1218e-02, -2.1404e-02, 8.2681e-01, -3.0264e-01, -7.6556e-02, 2.9048e-01, 3.4054e-01, 2.5881e-02, 1.9680e-01, 3.1750e-01, 4.4904e-01, -5.4009e-02, 3.5515e-01, -2.5754e-01, -3.0279e-01, -3.3378e-01, -3.1998e-01, 3.8379e-01, -1.5646e-01, 2.8057e-02, 1.3828e-01, 5.9443e-02, 1.5303e-01, 1.5804e-01, -2.3152e-01, 1.8590e-01, -1.7906e-01, -4.8659e-02, 3.5145e-01, 5.8912e-01, -1.8931e-01, 1.1240e-01, 8.6221e-02, -1.2742e-01, 1.1433e-02, -1.0033e-01, 1.9416e-01, 7.5975e-01, -8.4944e-03, -1.0347e-01, 3.3857e-01, -3.2450e-03, -2.5949e-01, -1.3049e-01, -2.7429e-01, 3.6076e-01, 2.2456e-02, 1.8395e-01, -1.1796e-02, -1.1933e+00, 1.9488e-01, -4.7571e-01, 8.2972e-02, 2.0567e-02, -5.6939e-01, 1.8543e-01, 1.5819e-01, -2.6344e-02, -1.3586e-01, -3.0559e-01, -9.1858e-02, 1.5596e+00, -1.1252e-01, -9.1376e-01, 1.4403e-01, 5.4933e-01, -8.1129e-03, -2.0602e-01, 2.7938e-01, -1.7516e-01, 3.3268e-01, -7.0229e-01, -1.8966e-01, 3.7934e-01, -1.4382e+00, 4.1491e-02, -7.4356e-01, 4.4283e-01, 2.2037e-01, 9.5358e-02, 1.8080e-01, -1.0913e-01, -6.1602e-02, -4.7201e-01, -1.2602e-01, -2.5680e-01, -1.3158e-01, -9.7970e-02, 3.1354e-01, 2.4643e-01, -3.5605e-02, 2.2407e-01, -5.6102e-01, -4.3781e-02, -6.6948e-02, 2.3365e-01, 2.0224e-01, -4.8406e-01, -1.8340e-01, 2.6515e-01, 3.5039e-02, 1.4085e-01, -3.1442e-01, 3.6208e-01, -4.2783e-01, 1.0023e-01, 5.7867e-01, 9.1382e-02, 7.7051e-01, 3.5109e-02, 2.1105e-01, 1.7448e-01, -8.4044e-03, 6.7720e-01, -4.8796e-01, -2.3340e-01, 1.2858e-01, -7.6849e-02, 2.9516e-01, 6.7951e-02, -2.0271e-01, -4.4297e-01, 3.9987e-01, 2.8524e-01, 9.0605e-03, 1.6964e-01, -2.3849e-01, -2.1865e-01, -5.8215e-02, -1.5098e-01, 3.9566e-02, -1.3141e-02, -4.2397e-01, -7.0880e-02, -2.0458e-01, 3.3506e-01, -2.1963e-01, 2.9304e-01, 3.5155e-01, -8.0974e-02, -2.9205e-02, 2.8120e-01, 7.4953e-01, -2.0307e-01, 2.9373e-01, 1.0246e-01, 2.5897e-01, 6.0300e-03, -1.3105e-01, 3.3122e-01, 3.9685e-01, -5.5882e-01, 2.2932e-01, 9.2667e-03, -2.8109e-01, -2.2059e-01, -1.4619e-02, -4.6908e-01, -4.6781e-01, -2.1942e-02, 9.2600e-02, -4.1449e-01, -2.8908e-01, -3.4762e-01, -9.6450e-02, 2.1795e-01, 2.1334e-01, -2.4707e-01, -1.4550e-01, -2.4918e-01, 2.2423e-01, -7.1012e-02, -7.5370e-02, -2.2632e-01, -2.5647e-01, -8.1591e-01, -3.4810e-01, 4.4667e-01, -1.4343e-01, 4.2052e-01, 7.8439e-02, -4.4730e-02, 3.6128e-01, -4.1957e-01, -1.6465e-01, 5.5771e-02, 3.9537e-02, -1.5157e-01, 1.5093e-01, -3.7800e-01, 9.5192e-02, 5.4860e-02, -1.2122e-01, -1.1731e-01, 1.5047e+00, 4.1262e-01, 3.6463e-01]), 'b53e76c0-9e5d-41a9-8c8b-014a13a61cfe': 'dog', 'e92d248d-8ea8-4e2b-8254-5988acfea072': 'canine'}