Embeddings with Sentence Transformers and Pinecone for Question Answering in Spanish

Experimenting with Pinecone vector database and Sentence Transformers and working with text data in Spanish, using Sentence Transformers to produce embeddings and then we create an index in Pinecone to store the embeddings. The goal is to build a question-answering model in Spanish.

Fine-tune a Llama 2 7B parameters model in 4-bit to generate Python Code

A notebook on how to instruction fine-tune the new Llama-2 7B model using PEFT, QLoRa and the Huggingface utilities. The objective task is to generate a Python coder that creates Python functions. The dataset contains 18,000 examples in alpaca format. It took about 3 hours to train in a A100 GPU on

Predicting mortgage approvals using Azure Machine Learning Studio

- Exploratory data analysis and visualization on data, extracting some insights and useful information - Apply some feature engineering and selection to improve our model - Design a boosted decision tree model on Azure Machine Learning Studio

Using Machine Learning to identify Accents in Spectrograms of Speech

- Study and analysis of spectrograms of speech - Development of a convolutional neural network to classify the accent of a speech - Building using Keras Tensorflow and Azure Machine Learning Services

Retrieval Augmented Generation (RAG): LLM-powered Q&A Chatbot to chat with your documents

Extend an LLM-powered Chatbot, retrieving and reranking a semantic search, to chat with a PDF file using Langchain and Pinecone vector database. Build a demo app with Streamlit and then publish it in Stream Community Cloud.

And end-to-end AWS Sagemaker Pipeline to prepare data, train, evaluate and conditional deployment of a Keras Text Classification model

Our goal is to define, build, and run a machine learning pipeline, end-to-end, that we can run to process data, build and train a keras model, evaluate the model on the validation dataset, and finally, if the performance or accuracy of the model is acceptable, deploy the model to an endpoint.

Train and Evaluate a Sentence Transformer model for news in Spanish in AWS Sagemaker

This project is a tutorial about how to train or finetune a sentence transformer model for Spanish language. For this exercise, we will use a very helpful library, SentenceTransformers is a Python framework for state-of-the-art sentence, text and image embeddings.