High Chunking Question 161 of 223

What is late chunking or contextual retrieval?

GenAI / LLM · Speak this in 60–90 seconds · Faridabad & Delhi NCR

PICTURE THIS: A SENTENCE BECOMES TOKENS

The model does not read letters like humans. It reads these pieces, then predicts the next one.

Simple meaning

Late chunking embeds a long passage with full context then pools token vectors into chunk embeddings so boundaries lose less meaning.

WHY — Tokens instead of words?

LLMs use tokens (not full words) because it helps them:

Chunking questions separate people

who only read docs from people who shipped.

Keep it short, concrete,

and tied to GenAI / LLM work.

Stable token IDs

Each piece maps to a number the network can learn.

Fits the model

Fixed pieces are what transformers expect as input.

STEPS — What happens with tokens?

Before the model can read a sentence, it goes through these steps:

1
Late chunking embeds a
long passage with full context then pools token vectors into chunk embeddings so boundaries lose less meaning.
2
Contextual retrieval prepends document
titles or LLM-generated context to each chunk before embedding.
3
Embeddings
Both attack the isolated-chunk problem.
4
Context mix
Attention looks at nearby tokens together.
5
Next token
The model scores what should come next.
6
Decode
IDs turn back into readable text.

EXAMPLE — See it in action

Let's see how a real sentence is tokenized (tokens may vary by model):

Input text

“Contextual retrieval prepends document titles or LLM-generated context to each c”

Tokenized output

Contextualretrievalprependsdocumenttitlesor

Token IDs (example)

2987408337471632900

Note: Actual tokens and IDs depend on the tokenizer (e.g., GPT, Llama, etc.).

Key takeaway

Late chunking embeds a long passage with full context then pools token vectors into chunk embeddings so boundaries lose less meaning. Contextual retrieval prepends document titles or LLM-generated context to each chunk before embedding.