Learning Dynamics of Meta-Learning in Small Model Pretraining
By: David Demitri Africa , Yuval Weiss , Paula Buttery and more
Potential Business Impact:
Trains AI faster and easier to understand.
Large language models are powerful but costly. We ask whether meta-learning can make the pretraining of small language models not only better but also more interpretable. We integrate first-order MAML with subset-masked LM pretraining, producing four LLama-style decoder-only models (11M-570M params), and evaluate it on a fundamental NLP task with many settings and real-world applications. Compared with vanilla training, our model (i) reaches the same loss up to 1.6x sooner, (ii) improves F1 on multilingual Universal NER under equal compute, and (iii) makes the training dynamics easy to read: first the network's representations fan out ("diversify") and later they collapse into a smaller, shared subspace ("compress"). This two-stage shift shows up as a rise-and-fall in both effective-rank curves and attention-head entropy. The same curves pinpoint which layers specialise earliest and which later reconverge, giving a compact, interpretable signature of meta-adaptation. Code, checkpoints and WandB logs are released.
Similar Papers
Meta-Pretraining for Zero-Shot Cross-Lingual Named Entity Recognition in Low-Resource Philippine Languages
Computation and Language
Teaches computers to understand new languages faster.
Domain-Adaptive Continued Pre-Training of Small Language Models
Computation and Language
Makes small AI smarter with less computer power.
EvoLM: In Search of Lost Language Model Training Dynamics
Computation and Language
Helps build better AI by testing its learning steps.