pablo-tech Goto Github PK

bayesian-structure-learning

Search of an optimal Bayesian Network, assessing its best fit to a dataset, via an objective scoring function. Created at Stanford University, by Pablo Rodriguez Bertorello

bert-legal-classification

CaseText Court Case analysis with fine-tuned BERT Transformer

challenge--featurecolumnforclassification

Columns in TensorFlow modeling: numeric, bucketized, categorical, embedding, hashed, crossed

challenge--predictshipcrewsize

conditioning-vs-performance-in-deep-neural-networks

Investigation of neural network conditioning under regularization approaches including Stochastic Gradient Descent. Research at Stanford University, by: Jakub Dworakowski, and Pablo Rodriguez Bertorello

customerlifetimevalue

Customer Lifetime Value estimation model

deep-learning-experiments

Notes and experiments to understand deep learning concepts

deepthinkingbattlesnake

google-t5flan-finetune

This repository contains code for extending the Stanford Alpaca synthetic instruction tuning to existing instruction-tuned models such as Flan-T5.

heapallocator

Enhancements on Bryant and O'Hallaron's Computer Systems

huffman-encoder-with-trees-heaps-hashes

Implementation of David Huffman's 1952 Minimal-Redundancy Codes algorithm, one of the most cited papers in Computer Science. By Pablo Rodriguez Bertorello at Stanford University

image-inventory-reconciliation-with-svm-and-cnn

Response to Amazon's Bin Image Data Set Challenge. Inventory reconciliation with machine learning: SVMs and CNNs. Research at Stanford University, by: Pablo Rodriguez Bertorello, Sravan Sripada, and Nutchapol Dendumrongsup

incubator-superset

Apache Superset is a Data Visualization and Data Exploration Platform

keras-io

Keras documentation, hosted live at keras.io

kmeans-initializationalgorithms-euclideanvsmanhattan

A comparison of Random. vs Far centroid initialization, with Euclidean vs Manhattan distance

latentfactorrecommendations

Instead of computing Singular Value Decomposition, which fits to no-rating as if zero-rating, machine learn rating matrix decomposition

naivebayesvslogisticregression

A comparison of machine learning algorithms: Naive Bayes vs Logistic Regression. Created at Stanford University, by Pablo Rodriguez Bertorello

nanogpt

The simplest, fastest repository for training/finetuning medium-sized GPTs.

nanogpt-colab

pagerank-vs-hubsauthorities

Comparison of Google's Page Rank vs Hubs and Authorities on the Internet

questeval

rapid-reinforcement-learning

The Courchevel environment eases the development of streaming Reinforcement Learning algorithms. Research at Stanford University, by Pablo Rodriguez Bertorello

reasoned-retrieval

Reasoning and Acting (ReAct) distillation from GPT4 to a small open source model

reinforcementlearning

Grids, mountains, and mysterious problems. Solved with Partially-Observable Markov Decision Procesees. Created at Stanford University, by Pablo Rodriguez Bertorello

smate--syntheticminorityadversarialtechnique

The novel SMate approach leverages GAN minority-class image generators, which benefit from Transfer Learning from majority-class image generators. Consequently, SMate outperforms SMOTE for imbalanced image data-sets. Research at Stanford University, by: Pablo Rodriguez Bertorello, Liang Ping Koh

textsimplification-tutorials

Sentence Simplification natural language algorithms

tigermarkovdecisionproblem