DrejcPesjak

Follow

Drejc Pesjak DrejcPesjak

Follow

Phd in LLM interpretability.

18 followers · 35 following

Achievements

Achievements

Highlights

Pro

DrejcPesjak/README.md

AI, in perspective

A small observatory for builders and researchers: which language models attract use, and where this week’s research is going.

Explore the models · Still image

Read the papers · Still image

_{Snapshots from Hugging Face · model downloads measure activity, not quality · papers follow HF daily-list dates · animations play once}

How it works · Label comparison · Design · Collection status

Pinned Loading

DPhate-double-paraphrasing-hate-speech DPhate-double-paraphrasing-hate-speech Public

Bachelor's thesis on removing hate from online comments using paraphrasing: algorithm DPhate

Python
scaling-monosemanticity-llama scaling-monosemanticity-llama Public

Reproducing Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet using LLaMA. This project explores monosemantic neurons in large language models, implementing and extend…

Jupyter Notebook 5
minigrid-crewai minigrid-crewai Public

Sense–Plan–Code–Act (SPCA) framework — LLM+PDDL planning, code generation and ROS2/UR5 simulation for embodied robotic manipulation.

Jupyter Notebook 6 1
unbalanced-media unbalanced-media Public

Analysis of Unbalanced Slovenian Media News Outlets - Left vs. Right Wing

Python 1
weather-prediction-mlops weather-prediction-mlops Public

ML in the cloud project for the universtiy course Cloud Computin (RSO)

Jupyter Notebook
nyc-violation-tickets-analysis nyc-violation-tickets-analysis Public

Analysis and prediction of NYC violation tickets using big data and machine learning techniques.

Jupyter Notebook