the only real way to write a thesis, with a bunch of functionality for custom _everything_

homepage

Whisper Subtitler: A simple one-command subtitle transcriber for input audio and video using whisper models.

Various dotfiles. MORE DOTS!

Guides and code for hosting your own local LLM through an API (such as ollama w/ llama.cpp) to access it by helpers, including tool use and data creation.

my ntnu homepage: https://folk.ntnu.no/tollefj/ mirrored at https://tollefj.github.io/public_html/

Code for the paper: "Margins in Contrastive Learning: Evaluating Multi-task Retrieval for Sentence Embeddings"

this package allows you to fetch the arguments of a function from a simple search string, matching a function within a module.

Code for the paper "Cross-Lingual Sentence Compression for Length-Constrained Subtitles in Low-Resource Settings"

Continued development of the system for the paper "Enhancing Criminal Investigation Analysis with Summarization and Memory-based Retrieval-Augmented Generation: A Comprehensive Evaluation of Real C…

UNER converted NorNE data

conversion and merging of NARC and UD

A simple inference server for llama cpp python, based on prompt configurations and more.

Course page for TDT4310 Intelligent Text Analytics and Language Understanding, spring 2024.

A collection of web scrapers

Advancing Knowledge Discoveries in Criminal Investigations with Semantic Textual Similarity

A lightweight sentence-pair encoder for similarity tasks. Includes a data augmentation pipeline from sentence-transformers.

Some llm examples for Norwegian

a simplified setup for basic python applications

A baseline for CorefUD