Tag
16 articles
Learn how to set up and use the Step 5 Preview open weights model from StepFun, a 600B parameter sparse Mixture-of-Experts model with 27B active parameters and 1M token context window.
Gigatoken, a new Rust-based BPE tokenizer, achieves text encoding speeds of up to 24.53 GB/s, outperforming HuggingFace tokenizers by 989x and tiktoken by 681x.
Learn to build a research agent framework similar to Perplexity's WANDR benchmark that evaluates AI systems' ability to discover evidence and provide verifiable sources.
Learn to build an AI-powered document processing system that mimics core Office functionality using Python, NLP libraries, and machine learning frameworks.
Learn to build and deploy an AI-powered sentiment analysis tool using OpenAI, Hugging Face transformers, and GPU acceleration - similar to technologies used by MANGOS companies.
Perplexity AI open-sources a new Unigram tokenizer that reduces p50 latency by 5x and cuts CPU utilization by 5-6x compared to Hugging Face tokenizers.
Learn to build a practical AI output validation system that combines multiple NLP techniques to assess the quality and reliability of AI-generated content, addressing the growing need for validating AI outputs in professional environments.
A tutorial from MarkTechPost demonstrates how to build knowledge graph generation pipelines using kg-gen, NetworkX, and interactive visualizations for extracting structured insights from text.
Learn to build a Python-based tool that detects potential AI-generated content in research papers, helping maintain academic integrity in the face of stricter arXiv policies.
Tokenization drift is a subtle but critical issue in AI where minor formatting differences in input text cause models to behave inconsistently. This phenomenon can lead to unexpected performance drops without any changes to data or logic.
Learn to build an AI-powered search query analyzer that can categorize and visualize search trends similar to Google's AI systems. This intermediate tutorial covers NLP preprocessing, TF-IDF vectorization, and clustering techniques.
This explainer explores BudouX, an advanced tool for intelligent multilingual text wrapping that uses NLP and machine learning to break lines at semantic boundaries, improving readability and user experience across different languages.
