/blog
Hey! 👋 I’m Mihai, a PhD student at Babeș-Bolyai University, specializing in Small Language Models (SLMs) for Romanian. My passion lies in Natural Language Processing (NLP) and Machine Learning (ML).
A blog about machine learning and natural language processing
In this blog, you’ll find a variety of posts ranging from in-depth tutorials on machine learning algorithms to the latest advancements in NLP. I also share insights from my research, including challenges faced and solutions discovered. Whether you’re a beginner or an expert, there’s something here for everyone.
🌐 Language
While most of the content on this blog will be in English to ensure it is accessible to a global audience, you’ll also find some articles written in 🇷🇴 Romanian. These are intended to support local developers, researchers, and enthusiasts who are particularly interested in Romanian NLP and its applications. My goal is to balance accessibility with relevance, making the blog useful both internationally and within Romania.
For more information, please visit my detailed about page. You can also connect with me on X, LinkedIn, and follow my work on GitHub.
/latest posts
-
The TinyFabulist Ecosystem: Connecting Generation, Translation, and Evaluation
-
Fine-Tuning Small Models for Diacritic Restoration
-
Designing Open-Weight Judge Panels for Text Evaluation
-
Running LLM Evaluation Locally with Ollama on Apple Silicon
-
TF3: Compact Romanian LMs on Synthetic Microfiction
-
Training a Small Language Model from Scratch
-
Presenting at InnoComp 2025: LLMs for Romanian Diacritics
-
One Year In: PhD Reflections and What Comes Next
-
The Romanian Diacritics Challenge: Why It Still Matters
-
Building Large-Scale EN-RO Translation Resources
-
LoRA Fine-Tuning for Literary Translation
-
Evaluating Translation Quality Beyond BLEU
-
Literary Translation with Open Models: The TF2 Project
-
LLMs as Judges: Surveying the Evaluation Landscape
-
TF1-EN-3M: My First Preprint and Dataset Release
-
Synthetic Data Generation with LLMs: A Survey Published in IEEE Access
-
Scaling Synthetic Fable Generation to 3 Million Stories
-
The TinyFabulist Idea: Structured Synthetic Narratives
-
NumPy vs Loops: Performance Analysis
-
Simple Perceptron Notebook
-
Navigating the NLP Literature Landscape
-
Starting a PhD in NLP: Research Directions and Early Decisions