Joel Niklaus's picture

Open to Collab

Joel Niklaus

joelniklaus

·

https://niklaus.ai

AI & ML interests

Pretraining, Instruction Tuning, Domain Adaptation, Benchmarks, Legal Datasets

Recent Activity

new activity 1 day ago

joelniklaus/lextreme:Expanding Ukrainian legal tasks in LEXTREME + feedback request

new activity 1 day ago

joelniklaus/lextreme:CPT on 161B tokens of Ukrainian court decisions — interested in updates?

updated a dataset 2 days ago

LEXam-Benchmark/LEXam

View all activity

Organizations

upvoted a collection 3 months ago

🤏 Smol-Data

Tried and tested mixes for strong pretraining. Inspired by https://huggingface.co/blog/codelion/optimal-dataset-mixing • 14 items • Updated Mar 2 • 12

upvoted an article 3 months ago

Article

Compute and Competition in AI: Different FlOPs for Different Folks

sasha

•

Feb 12

• 15

upvoted an article 4 months ago

Article

Introducing Trackio: A Lightweight Experiment Tracking Library from Hugging Face

+3

abidlabs, znation, nouamanetazi, sasha, qgallouedec

•

Jul 29, 2025

• 223

upvoted a paper 4 months ago

Recursive Language Models

Paper • 2512.24601 • Published Dec 31, 2025 • 96

upvoted a collection 9 months ago

Common Pile v0.1

All resources related to Common Pile v0.1, an 8TB dataset of public domain and openly licensed text • 4 items • Updated Jun 6, 2025 • 40