note

Building My Own AI (GPT) #5 Starting Marin v1.0: Creating a Custom BPE Tokenizer

In the previous post (#4), I confirmed that I had reached the stage just before "gibberish that looks like text." I named this model Marin v0.1 and created a roadmap for what follows. Starting from ...
GitHub

SMILES Pair Encoding: A Data-Driven Substructure Tokenization Algorithm for Deep Learning

SMILES Pair Encoding (JCIM) first learns a vocabulary of high frequency SMILES substrings from a large chemical dataset (e.g., ChEMBL) and then tokenizes SMILES based on the learned vocabulary for ...
Scientific Research Publishing

Deep Reinforcement Learning for Phishing Detection with Transformer-Based Semantic Features ()

Phishing is a form of cybercrime in which people are deceived into exposing their personal information which can result in financial loss. These attacks are often executed via fraudulent messages, ...
GitHub

Full Sequence-to-Sequence Encoder-Decoder Transformer for Neural Machine Translation

This repository contains a mathematically precise implementation of the original "Attention Is All You Need" Transformer architecture. Built entirely from scratch, the model maps English source ...
IEEE

Electrical and Computer Engineering Researches (ICECER), International Conference on

Data centers are deemed essential infrastructures for the digital economy and a wide variety of operational activities such as cloud computing & Artificial Intelligence, big data analytics, e-commerce ...
Frontiers

DiscoVerse: multi-agent pharmaceutical co-scientist for traceable drug discovery and reverse translation

We assess DiscoVerse through expert evaluation of source-linked outputs. Our evaluation spans a selected subset of 180 molecules from Roche's research and development repositories, encompassing over 0 ...