MORE FLOPS • FEWER REGRETS • MORE FLOPS • FEWER REGRETS • → AI RESEARCH → NEURAL NETS → → AI RESEARCH → NEURAL NETS →
→
→
→
→

AI Research Engineer @ Aleph Alpha

Max Meuer

Experience

Aleph Alpha

2023 —
  • →pre-training efficiency for hierarchical autoregressive transformers
  • →7B multilingual embedding model training
  • →synthetic data generation for post-training

Working Student

Merck KGaA

Cloud infrastructure

2022

Campana & Schott

Entity resolution

2021 — 22

Lufthansa

Visual recognition

2019 — 21

Publications

TFree-HAT-7B

Hierarchical autoregressive transformer

Model Release

Pharia-1-Embedding

Instruction-tuned embedding modelrelo

Model Release

Divergent Token Metrics

Quantization degradation & pruning

NAACL 2024

Stack

PythonTritonPyTorchMegatron-LMDockerPySpark

Education

M.Sc. Computer Science

TU Darmstadt · 2023

B.Sc. Information Systems

TU Darmstadt · 2020