update package versions for medical device training assistant (#264)

* update package versions

* update README

Author: jin-nvidia

industries/healthcare/medical-device-training-assistant/README.md

@@ -51,7 +51,10 @@ Complete the [common prerequisites](../../../docs/common-prerequisites.md). This
 Assuming you would like to enable RIVA ASR and TTS to use your voice for interacting with the RAG, refer to the [Local RIVA Server](../../../docs/riva-asr-tts.md#local-riva-server) section on running RIVA containers on the local machine's compute resources, or refer to [Riva API Endpoint on NVIDIA API Catalog](../../../docs/riva-asr-tts.md#riva-api-endpoint-on-nvidia-api-catalog) to access RIVA speech models on NVIDIA API Catalog instead of hosting them locally. Refer to the rest of the [documentation on RIVA](../../../docs/riva-asr-tts.md) to configure the docker-compose.yaml file. 
 
 ### Hardware Requirement
-The Milvus Vector Database is GPU-enabled by default in this IFU RAG application, which is specified in [docker-compose-vectordb.yaml](./docker-compose-vectordb.yaml). Please refer to the section [Configuring Milvus with GPU Acceleration](../../../docs/vector-database.md#configuring-milvus-with-gpu-acceleration) to see how to set GPU acceleration on and off.
+There is no GPU requirement to run this repository, and you could optionally configure components of this repository to utilize GPUs.
+
+You can optionally self host the NIMS for [LLM](https://docs.nvidia.com/nim/large-language-models/latest/getting-started.html), [embedding](https://docs.nvidia.com/nim/nemo-retriever/text-embedding/latest/getting-started.html), and [reranking](https://docs.nvidia.com/nim/nemo-retriever/text-reranking/latest/getting-started.html) models, or use NVIDIA AI Endpoints. For RIVA ASR and TTS, as stated in [RIVA Setup](#riva-setup) you could choose to host them locally or use the endpoints.
+The Milvus Vector Database is CPU-enabled by default in this IFU RAG application, which is specified in [docker-compose-vectordb.yaml](./docker-compose-vectordb.yaml). Please refer to the section [Configuring Milvus with GPU Acceleration](../../../docs/vector-database.md#configuring-milvus-with-gpu-acceleration) to see how to set GPU acceleration on.
 
 ## Build and Start the Containers
 
@@ -125,7 +128,7 @@ If you see any errors while navigating the web UI, you could use `docker logs ch
 
 ## Next Steps
 
-- [Vector Database Customizations](../../../docs/vector-database.md) to disable GPU acceleration for the Milvus vector database.
+- [Vector Database Customizations](../../../docs/vector-database.md) to customize GPU acceleration for the Milvus vector database.
 - Stop the containers by running `docker compose down`.
 - If you're interested in adapting this application to use on premises machines for NVIDIA Inference Microservices (NIMs), for the LLM, embedding model and reranking model, please refer to [local-deploy README](../../../RAG/examples/local_deploy/README.md) for the requirement and configuration.
 

industries/healthcare/medical-device-training-assistant/src/chain_server/Dockerfile

@@ -8,7 +8,7 @@ ENV DEBIAN_FRONTEND noninteractive
 
 # Install required ubuntu packages for setting up python 3.10
 RUN apt update && \
-    apt install -y curl software-properties-common libgl1 libglib2.0-0 && \
+apt install -y curl software-properties-common libgl1 libglib2.0-0 libmagic1 libmagic-dev && \
     add-apt-repository ppa:deadsnakes/ppa && \
     apt update && apt install -y python3.10 && \
     apt-get clean
@@ -28,7 +28,7 @@ RUN --mount=type=bind,source=src/chain_server/requirements.txt,target=/opt/requi
 # TODO Maybe copy app
 ARG APP_PATH
 COPY ${APP_PATH} /opt/${APP_PATH}
-RUN python3.10 -m nltk.downloader averaged_perceptron_tagger
+RUN python3.10 -m nltk.downloader averaged_perceptron_tagger_eng
 
 # Copy required common modules for all examples
 # TODO copy only relevant files

industries/healthcare/medical-device-training-assistant/src/chain_server/requirements.txt

@@ -1,24 +1,25 @@
 fastapi==0.110.0
 uvicorn[standard]==0.27.1
 python-multipart==0.0.9
-langchain==0.1.9
+langchain==0.3.14
+langchain-community==0.3.14
+langchain-core==0.3.29
 unstructured[all-docs]==0.12.5
 sentence-transformers==3.0.0
-llama-index-core==0.10.27
-llama-index-readers-file==0.1.22
-llama-index-llms-langchain==0.1.3
-llama-index-embeddings-langchain==0.1.2
-llama-index-vector-stores-milvus==0.1.6
-llama-index-vector-stores-postgres==0.1.5
-pymilvus==2.4.0
+llama-index-core==0.12.10
+llama-index-readers-file==0.4.3
+llama-index-llms-langchain==0.5.0
+llama-index-embeddings-langchain==0.3.0
+llama-index-vector-stores-milvus==0.5.0
+llama-index-vector-stores-postgres==0.4.1
+pymilvus==2.5.3
 dataclass-wizard==0.22.3
 opencv-python==4.8.0.74
 minio==7.2.5
 asyncpg==0.29.0
 psycopg2-binary==2.9.9
-pgvector==0.2.5
-langchain-core==0.1.29
-langchain-nvidia-ai-endpoints==0.1.6
+pgvector==0.3.6
+langchain-nvidia-ai-endpoints==0.3.7
 opentelemetry-sdk==1.23.0
 opentelemetry-api==1.23.0
 opentelemetry-exporter-otlp-proto-grpc==1.23.0

industries/healthcare/medical-device-training-assistant/src/rag_playground/Dockerfile

@@ -8,7 +8,7 @@ ENV DEBIAN_FRONTEND noninteractive
 
 # Install required ubuntu packages for setting up python 3.10
 RUN apt update && \
-    apt install -y dpkg openssl libgl1 linux-libc-dev libksba8 curl software-properties-common build-essential libssl-dev libffi-dev && \
+    apt install -y dpkg openssl libgl1 linux-libc-dev libksba8 curl software-properties-common build-essential libssl-dev libffi-dev ffmpeg && \
     add-apt-repository ppa:deadsnakes/ppa && \
     apt update && apt install -y python3.10 python3.10-dev python3.10-distutils