All environments require additional Conda packages which can be installed with either the conda/environments/all_cuda-121_arch-x86_64.yaml or conda/environments/examples_cuda-121_arch-x86_64.yaml environment files.
Environment |
Supported |
Notes |
|---|---|---|
| Conda | ✔ | |
| Morpheus Docker Container | ✔ | Requires launching Triton and Milvus on the host |
| Morpheus Release Container | ✔ | Requires launching Triton and Milvus on the host |
| Dev Container | ✘ |
Purpose
The primary objective of this example is to demonstrate the construction of a pipeline that performs the following tasks:
Accepts a collection of documents.
Divides the documents into smaller segments or chunks.
Computes the embedding vector for each of these chunks.
Uploads the text chunks along with their associated embeddings to a Vector Database (VDB).
Source Documents
The pipeline is designed to process text-based input from various document types. Possible use cases could include structured documents like PDFs, dynamic sources such as web pages, and image-based documents through future Optical Character Recognition (OCR) integration.
For this demonstration, the source documents are obtained from RSS feeds combined with a web scraper. The rationale behind this selection includes:
Emulating practical cyber scenarios: Cybersecurity RSS feeds can serve as the foundation for a comprehensive knowledge database, such as for a security chatbot.
Minimizing external dependencies: Relying on RSS feeds and web scraping avoids the need for specialized datasets or API keys.
Embedding Model
The pipeline can accommodate various embedding models that transform text into vectors of floating-point numbers. Several models from Huggingface, such as
paraphrase-multilingual-mpnet-base-v2,e5-large-v2, andall-mpnet-base-v2, have been evaluated for compatibility.For the purposes of this demonstration, the model
all-MiniLM-L6-v2is employed. This model is included via LFS in this repository, and was chosen for its efficiency and compactness, characterized by a smaller embedding dimension of 384.
Vector Database Service
The architecture is agnostic to the choice of Vector Database (VDB) for storing embeddings and their metadata. While the present implementation employs Milvus due to its GPU-accelerated indices, the design supports easy integration with other databases like Chroma or FAISS, should the need arise.
Implementation Details
The pipeline is composed of three primary components:
Document Source Handler: This component is responsible for acquiring and preprocessing the text data. Given that we are using RSS feeds and a web scraper in this example, the handler’s function is to fetch the latest updates from the feeds, perform preliminary data cleaning, and standardize the format for subsequent steps.
Embedding Generator: This is the heart of the pipeline, which takes the preprocessed text chunks and computes their embeddings. Leveraging the model
all-MiniLM-L6-v2from Huggingface, the text data is transformed into embeddings with a dimension of 384.Vector Database Uploader: Post embedding generation, this module takes the embeddings alongside their associated metadata and pushes them to a Vector Database (VDB). For our implementation, Milvus, a GPU-accelerated vector database, has been chosen.
Rationale Behind Design Decisions
The selection of specific components and models was influenced by several factors:
Document Source Choice: RSS feeds and web scraping offer a dynamic and continuously updating source of data. For the use-case of building a repository for a cybersecurity, real-time information fetching is a reasonable choice.
Model Selection for Embeddings:
all-MiniLM-L6-v2was chosen due to its efficiency in generating embeddings. Its smaller dimension ensures quick computations without compromising the quality of embeddings.Vector Database: For the purposes of this pipeline, Milvus was chosen due to its popularity, ease of use, and availability.
Prerequisites
Before running the pipeline, we need to ensure that the following services are running:
Ensure LFS files are downloaded
To retrieve models from LFS run the following:
./scripts/fetch_data.py fetch models
Milvus Service
Follow the instructions here to install and run a Milvus service.
Triton Service
Pull the Docker image for Triton:
docker pull nvcr.io/nvidia/tritonserver:23.06-py3
From the Morpheus repo root directory, run the following to launch Triton and load the
all-MiniLM-L6-v2model:docker run --rm -ti --gpus=all -p8000:8000 -p8001:8001 -p8002:8002 -v $PWD/models:/models nvcr.io/nvidia/tritonserver:23.06-py3 tritonserver --model-repository=/models/triton-model-repo --exit-on-error=false --model-control-mode=explicit --load-model all-MiniLM-L6-v2
This will launch Triton and only load the
all-MiniLM-L6-v2model. Once Triton has loaded the model, the following will be displayed:+------------------+---------+--------+ | Model | Version | Status | +------------------+---------+--------+ | all-MiniLM-L6-v2 | 1 | READY | +------------------+---------+--------+
Running the Morpheus Pipeline
The top-level entry point for each of the LLM example pipelines is examples/llm/main.py. This script accepts a set of options and a pipeline to run. For the purposes of this document, we’ll focus on the vdb_upload pipeline option, which incorporates various functionalities like handling RSS and filesystem sources, embedding configurations, and vector database (VDB) settings.
Configuration Balance Considerations
When configuring the Morpheus Pipeline, especially for stages like the RSS source and the Vector Database Upload, it’s important to balance responsiveness and performance.
RSS Source Stage: The RSS source stage is responsible for yielding webpage links for processing. A larger batch size at this stage can lead to decreased responsiveness, as the subsequent web scraper stage may take a considerable amount of time to retrieve and process all the items in each batch. To ensure a responsive experience for users, it’s recommended to configure the RSS source stage with a relatively smaller batch size. This adjustment tends to have minimal impact on overall performance while significantly improving the time to process each batch of links.
Vector Database Upload Stage: At the other end of the pipeline, the Vector Database Upload stage has its own considerations. This stage experiences a significant transaction overhead. To mitigate this, it is advisable to configure this stage with the largest batch size possible. This approach helps in efficiently managing transaction overheads and improves the throughput of the pipeline, especially when dealing with large volumes of data.
Balancing these configurations ensures that the pipeline runs efficiently, with optimized responsiveness at the RSS source stage and improved throughput at the Vector Database Upload stage.
Run example:
Default example usage, with pre-defined RSS source
python examples/llm/main.py vdb_upload pipeline \
--enable_cache \
--enable_monitors \
--embedding_model_name all-MiniLM-L6-v2
Usage with CLI-Defined Sources:
Example: Defining an RSS Source via CLI
python examples/llm/main.py vdb_upload pipeline \
--source_type rss \
--interval_secs 300 \
--rss_request_timeout_sec 5.0 \
--enable_cache \
--enable_monitors \
--embedding_model_name all-MiniLM-L6-v2
Example: Defining a Filesystem Source via CLI
python examples/llm/main.py vdb_upload pipeline \
--source_type filesystem \
--file_source "./morpheus/data/*" \
--enable_monitors \
--embedding_model_name all-MiniLM-L6-v2
Example: Combining RSS and Filesystem Sources via CLI
python examples/llm/main.py vdb_upload pipeline \
--source_type rss --source_type filesystem \
--file_source "./morpheus/data/*" \
--interval_secs 600 \
--enable_cache \
--enable_monitors \
--embedding_model_name all-MiniLM-L6-v2
The vdb_upload command has its own set of options and commands:
Commands:
export-triton-modellangchainpipeline
Exporting and Deploying a Different Model from Huggingface
If you’re looking to incorporate a different embedding model from Huggingface into the pipeline, follow the steps below
using sentence-transformers/paraphrase-multilingual-mpnet-base-v2 as an example:
Identify the Desired Model:
Head over to the Huggingface Model Hub and search for the model you want. For this example, we are looking at
e5-large-v2.
Run the Pipeline Call with the Chosen Model:
Execute the following command with the model name you’ve identified:
python examples/llm/main.py vdb_upload export-triton-model --model_name \ sentence-transformers/paraphrase-multilingual-mpnet-base-v2 --triton_repo ./models/triton-model-repo
Handling Unauthorized Errors:
Please ensure you provide the correct model name. A common pitfall is encountering an
unauthorized error. If you see the following error:requests.exceptions.HTTPError: 401 Client Error: Unauthorized for url:
This typically means the model name you provided does not match the one available on Huggingface. Double-check the model name and try again.
Confirm Successful Model Export:
After running the command, ensure that the specified
--triton_repodirectory now contains the exported model in the correct format, ready for deployment.
$ ls ${MORPHEUS_ROOT}/models/triton-model-repo | grep paraphrase-multilingual-mpnet-base-v2 sentence-transformers/paraphrase-multilingual-mpnet-base-v2
Deploy the Model:
Reload the docker container, specifying that we also need to load paraphrase-multilingual-mpnet-base-v2
docker run --rm -ti --gpus=all -p8000:8000 -p8001:8001 -p8002:8002 \ -v $PWD/models:/models nvcr.io/nvidia/tritonserver:23.06-py3 tritonserver \ --model-repository=/models/triton-model-repo --exit-on-error=false --model-control-mode=explicit --load-model \ all-MiniLM-L6-v2 --load-model sentence-transformers/paraphrase-multilingual-mpnet-base-v2
You should see something similar to the following, indicating Triton has successfully loaded the model:
+----------------------------------+------------------------------------------------------------------------------------------+ | Option | Value | +----------------------------------+------------------------------------------------------------------------------------------+ | server_id | triton | | server_version | 2.35.0 | | server_extensions | classification sequence model_repository ... schedule_policy | | model_repository_path[0] | /models/triton-model-repo | | model_control_mode | MODE_EXPLICIT | | startup_models_0 | all-MiniLM-L6-v2 | | startup_models_1 | sentence-transformers/paraphrase-multilingual-mpnet-base-v2 | | strict_model_config | 0 | | rate_limit | OFF | | pinned_memory_pool_byte_size | 268435456 | | cuda_memory_pool_byte_size{0} | 67108864 | | cuda_memory_pool_byte_size{1} | 67108864 | | min_supported_compute_capability | 6.0 | | strict_readiness | 1 | | exit_timeout | 30 | | cache_enabled | 0 | +----------------------------------+------------------------------------------------------------------------------------------+
Update the Pipeline Call:
Now that the model has been exported and deployed, we can update the pipeline call to use the new model:
python examples/llm/main.py vdb_upload pipeline --model_name \ sentence-transformers/paraphrase-multilingual-mpnet-base-v2
Running the Langchain Pipeline (Optional)
Optional guide for running the Langchain pipeline, if applicable.## Developer Docs
A link to the developer documentation where the README.md is also linked.
Note: This pipeline will, by default, run continuously repeatedly polling the configured RSS sources. To run for a fixed number of iterations, add the --stop_after=N flag.