Frontier Lab Notes

Data Tokenization and Pretraining Objective

Architecture is only half the story. Frontier LLM behavior is strongly shaped by data, tokenization, objective, filtering, mixture design, and training curriculum.

Next-Token Prediction

Most decoder-only LLMs are pretrained with causal language modeling:

maximize log P(x_t | x_1, ..., x_{t-1})

Or equivalently, minimize cross-entropy loss over next-token predictions.

Why this works:

Why it is not enough:

This motivates Post Training Alignment and Reasoning.

Tokenization

Tokenization maps text into discrete ids. Common approaches use subword tokens such as BPE or unigram tokenization.

Tokenization affects:

A bad tokenizer can make some languages or domains more expensive by requiring more tokens for the same content.

Dataset Mixture

Frontier pretraining data is a mixture:

Mixture design decides what the model gets good at.

Data Quality

Important operations:

Data curation can beat naive scale.

Curriculum And Stages

Training can happen in stages:

Each stage changes behavior and can trade off capabilities.

Evaluation Risk: Contamination

If benchmark examples appear in training data, evaluation can overestimate capability.

Good research asks:

Research Questions

Related