doc changes + minor pkg additions

applications/GTE_mlx_RAG/README.md

@@ -0,0 +1,40 @@
+
+## CLI RAG application using GTE MLX (Apple silicon) and Lancedb
+This is a Command Line Interface app designed to provide users with quick and accurate responses to their queries based on the input file (pdf) via RAG architecture.
+
+## Overview
+It is built using `lancedb.embeddings.gte` embedding function which uses General Text Embeddings (GTE) model to embed documents, we have added support for Apple silicon devices by adding the MLX format of the model which can be accessed by the 'mlx=True' argument in the function. This app uses `mlx_lm` as the LLM for generating the final response.
+
+- ingest_pdf.py - Extracts text from input pdf and stores in vectorDB.
+- query_gte.py - Retrieves context relevent to question from vectorDB and augments prompt to generate RAG response.
+
+
+## Getting started
+
+* Install requirements
+
+Please install lancedb via git instead of PyPI as some latest features might be missing, to get the latest code, run the following in your virtual env :
+```bash
+pip install -e "git+https://github.com/lancedb/lancedb.git#egg=lancedb&subdirectory=python" 
+
+python3 -m pip install -r requirements.txt
+```
+
+* Create vectors from a pdf file and store in lancedb
+
+Store your input data in './data/' folder and run the following to ingest the vectors :
+```bash
+python3 ingest_pdf.py --pdf ./data/mossformer.pdf
+```
+
+* Query database and generate response
+
+Run the following command by adding your query after the '--question' argument to get the response from the llm
+```bash
+python3 query_gte.py --question "Explain 2.2.- Hybrid MossFormer and Recurrent Modules in simpler language."
+```
+
+
+## Sample response 
+
+![image](../../assets/GTE.png)

applications/GTE_mlx_RAG/requirements.txt

@@ -1,3 +1,4 @@
 unstructured[pdf]
 mlx
-mlx_lm
+mlx_lm
+ai2-olmo