Synthetic Pre-pretraining Survives Scale, but Not as a Grammatical Prior

arXiv:2609.39827 · cs.CL, cs.AI, cs.LG · Submitted 2026-09-30 · Read on arXiv

cs.CL, cs.AI, cs.LG

Submitted: 2026-09-30

Updated: 2026-09-30

Code: https://github.com/gucci-j/verify-ppt-at-scalehttps:

Terminology

• Benchmark — 11× in this paper · explained in 2 episodes
• Core — 11× in this paper · explained in 2 episodes
• Accuracy — 9× in this paper · explained in 1 episode
• Transformer — 9× in this paper · explained in 2 episodes
• Language Modeling — 6× in this paper
• Ablation — 5× in this paper · explained in 1 episode
• Empirical Methods in Natural Language Processing — 4× in this paper
• Abstraction — 3× in this paper · explained in 1 episode
• Commonsense Reasoning — 3× in this paper
• Reading Comprehension — 3× in this paper
• Adam — 2× in this paper · explained in 2 episodes
• EMMA — 2× in this paper · explained in 2 episodes
• Tokenization — 2× in this paper · explained in 2 episodes
• Alignment — 1× in this paper · explained in 1 episode
• Causal Reasoning — 1× in this paper · explained in 2 episodes
• Conference on Computational Natural Language Learning — 1× in this paper
• Fine-Tuning — 1× in this paper · explained in 5 episodes
• Foundation Model — 1× in this paper · explained in 3 episodes
• Generalizability — 1× in this paper · explained in 6 episodes
• Resource Constraints — 1× in this paper · explained in 3 episodes
• Robustness — 1× in this paper · explained in 5 episodes
• Scaling Laws — 1× in this paper · explained in 4 episodes
• Transfer Learning — 1× in this paper · explained in 10 episodes

Sources

• Logic Before Language: Pre-pretraining on Formal Derivations Fosters Skill Acquisition and Compressibility
• Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge
• Gemma 4 Technical Report
• Synthetic Pre-Pre-Training Improves Language Model Robustness to Noisy Pre-Training Data
• Code Pretraining Improves Entity Tracking Abilities of Language Models
• Kimi K3: Open Frontier Intelligence
• Training Language Models via Neural Cellular Automata
• StarCoder 2 and The Stack v2: The Next Generation
• EuroLLM-9B: Technical Report
• An Empirical Model of Large-Batch Training
• Qwen2.5 Technical Report
• Qwen3.5-Omni Technical Report
• Olmo 3
• Llama 2: Open Foundation and Fine-Tuned Chat Models
• MultiHashFormer: Hash-based Generative Language Models
• ReCoRD: Bridging the Gap between Human and Machine Commonsense Reading Comprehension

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering