Download and configure model on Docker image build

Why these changes are being introduced:

We have opted to include the model weights and assets inside the Docker image.  When the model is
small-ish like ours, this avoids the need to save the model to S3 and download each time the CLI is
invoked.

How this addresses that need:

The CLI command `download-model` is used within the Dockerfile itself to download the model.  As noted
via inline comments, the env vars `TE_MODEL_URI` and `TE_MODEL_DOWNLOAD_PATH` are also set in the Dockerfile.
This has a dual purpose.  First, these env vars inform the `download-model` CLI invocation within the
Dockerfile.  Second, they persist to the container and establish the model as default for all calls.

Side effects of this change:
* On Docker image build, the model will be downloaded from HuggingFace, configured, and included
in the final Docker image.

Relevant ticket(s):
* https://mitlibraries.atlassian.net/browse/USE-113

Author: ghukill

Dockerfile

@@ -18,4 +18,13 @@ COPY embeddings ./embeddings
 # Install package into system python, includes "marimo-launcher" script
 RUN uv pip install --system .
 
-ENTRYPOINT ["embeddings"]
+# Download the model and include in the Docker image
+# NOTE: The env vars "TE_MODEL_URI" and "TE_MODEL_DOWNLOAD_PATH" are set here to support
+#  the downloading of the model into this image build, but persist in the container and
+#  effectively also set this as the default model.
+ENV HF_HUB_DISABLE_PROGRESS_BARS=true
+ENV TE_MODEL_URI=opensearch-project/opensearch-neural-sparse-encoding-doc-v3-gte
+ENV TE_MODEL_DOWNLOAD_PATH=/model
+RUN python -m embeddings.cli --verbose download-model
+
+ENTRYPOINT ["python", "-m", "embeddings.cli"]

README.md

@@ -28,6 +28,20 @@ TE_MODEL_DOWNLOAD_PATH=# Download location for model
 HF_HUB_DISABLE_PROGRESS_BARS=#boolean to use progress bars for HuggingFace model downloads; defaults to 'true' in deployed contexts
 ```
 
+## Configuring an Embedding Model
+
+This CLI application is designed to create embeddings for input texts.  To do this, a pre-trained model must be identified and configured for use.  
+
+To this end, there is a base embedding class `BaseEmbeddingModel` that is designed to be extended and customized for a particular embedding model.
+
+Once an embedding class has been created, the preferred approach is to set env vars `TE_MODEL_URI` and `TE_MODEL_DOWNLOAD_PATH` directly in the `Dockerfile` to a) download a local snapshot of the model during image build, and b) set this model as the default for the CLI.
+
+This allows invoking the CLI without specifying a model URI or local location, allowing this model to serve as the default, e.g.:
+
+```shell
+uv run --env-file .env embeddings test-model-load
+```
+
 ## CLI Commands
 
 For local development, all CLI commands should be invoked with the following format to pickup environment variables from `.env`: