---
metadata:
  - name: generator
    content: Diplodoc Platform v5.52.0
alternate:
  - https://catboost.ai/docs/en/concepts/python-reference_dictionary_apply.md
  - href: en/concepts/python-reference_dictionary_apply.md
    type: text/markdown
    title: Markdown version
  - href: ../llms.txt
    type: text/markdown
    title: llms.txt
---
> **Documentation Index:** Fetch the complete configuration index at https://catboost.ai/docs/en/llms.txt

# apply

<!-- source: en/_includes/work_src/reusage-tokenizer/apply.md -->
Apply a previously trained dictionary to the input text.
<!-- endsource: en/_includes/work_src/reusage-tokenizer/apply.md -->


## Method call format {#call-format}

```
apply(data,
      tokenizer=None,
      unknown_token_policy=None)
```

## Parameters {#parameters}

### data

#### Description

The input text to apply the dictionary to.

A zero-, one- or two-dimensional array-like data.

**Data types**

string, numpy.ndarray, pandas.DataFrame

**Default value**

Obligatory parameter

### tokenizer

#### Description

The tokenizer for text processing.

If this parameter is specified and a one-dimensional data is input, each element in this list is considered a sentence and is tokenized.

**Data types**

[Tokenizer](https://catboost.ai/docs/en/concepts/python-reference_tokenizer.md)

**Default value**

None (the input data is considered tokenized)

### unknown_token_policy

#### Description

The policy for processing unknown tokens.

Possible values:
- Skip — All unknown tokens are skipped from the resulting token ids list (empty values are put in compliance)
- Insert — A coinciding ID is put in compliance with all unknown tokens. This ID matches the number of the tokens in the dictionary.

**Data types**

string

**Default value**

Skip

## Type of return value {#return-value}

A one- or two-dimensional array with token IDs.

## Example {#example}

```python
from catboost.text_processing import Dictionary

dictionary = Dictionary(occurence_lower_bound=0)\
    .fit(["his", "tender", "heir", "whatever"])

applied_model = dictionary.apply(["might", "bear", "his", "memory"])

print(applied_model)
```

Output:

```bash
[[], [], [1], []]
```

## An example with input string tokenization {#tokenizing-example}

```python
from catboost.text_processing import Dictionary, Tokenizer

tokenized = Tokenizer()

dictionary = Dictionary(occurence_lower_bound=0)\
    .fit(["his tender heir whatever"], tokenizer=tokenized)

applied_model = dictionary.apply(["might", "bear", "his", "memory"])

print(applied_model)
```

Output:

```bash
[[], [], [1], []]
```