trl/pr_5607/en/chat_template_utils.md

# Chat template utilities

## clone_chat_template[[trl.clone_chat_template]]

#### trl.clone_chat_template[[trl.clone_chat_template]]

[Source](https://github.com/huggingface/trl/blob/vr_5607/trl/chat_template_utils.py#L27)

Clones a chat template from a source tokenizer to the target tokenizer and updates the model accordingly.

This function:
- Copies the chat template from a source tokenizer to the target tokenizer.
- Adds any new tokens from the source tokenizer to the target tokenizer.
- Sets and synchronizes the EOS token across the tokenizer and model.
- Resizes the model's token embeddings to match the new vocabulary size, optionally rounding it up to a multiple of
  a specified value. In such cases, dummy tokens are added to the tokenizer to ensure the vocabulary size matches
  the embedding dimensions.

Example:
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
from trl import clone_chat_template

model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-3.2-1B")
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-3.2-1B")
model, tokenizer, added_tokens = clone_chat_template(model, tokenizer, "Qwen/Qwen3-0.6B")
```

**Parameters:**

model ([PreTrainedModel](https://huggingface.co/docs/transformers/main/en/main_classes/model#transformers.PreTrainedModel)) : Model to update.

tokenizer (`PreTrainedTokenizer`) : Tokenizer to update.

source_tokenizer_path (`str`) : Path or identifier of the pretrained tokenizer to clone from.

resize_to_multiple_of (`int` or `None`, *optional*, defaults to `64`) : The embedding layer will be resized to the new vocabulary size. If this is not `None`, it will round up the new vocabulary size to the nearest multiple of this value.

**Returns:**

`model ([PreTrainedModel](https://huggingface.co/docs/transformers/main/en/main_classes/model#transformers.PreTrainedModel))`

Updated model with resized token embeddings and EOS token configured.
tokenizer (`PreTrainedTokenizer`):
Updated tokenizer with the chat template and special tokens applied.
added_tokens (`list[int]`):
List of tokens that were added to the tokenizer from the source tokenizer.

## is_chat_template_prefix_preserving[[trl.chat_template_utils.is_chat_template_prefix_preserving]]

#### trl.chat_template_utils.is_chat_template_prefix_preserving[[trl.chat_template_utils.is_chat_template_prefix_preserving]]

[Source](https://github.com/huggingface/trl/blob/vr_5607/trl/chat_template_utils.py#L453)

Check whether the chat template preserves prefixes when applied.

A prefix-preserving chat template renders earlier messages identically regardless of what messages follow. This
property is required by `_get_tool_suffix_ids`, which extracts tool response formatting tokens by comparing
tokenizations with and without tool messages appended.

**Parameters:**

processing_class (`PreTrainedTokenizer` or `ProcessorMixin`) : Tokenizer or processor instance to check.

**Returns:**

``bool``

`True` if the chat template preserves prefixes, `False` otherwise.

## get_training_chat_template[[trl.get_training_chat_template]]

#### trl.get_training_chat_template[[trl.get_training_chat_template]]

[Source](https://github.com/huggingface/trl/blob/vr_5607/trl/chat_template_utils.py#L515)

Get a training-compatible chat template, if needed.

Returns a patched chat template that is prefix-preserving and includes `{%% generation %%}` / `{%% endgeneration
%%}` markers for assistant-only loss masking. Returns `None` if the tokenizer's template already satisfies both
requirements. Currently DeepSeek-V3, GPT-OSS, LLaMA 3, Qwen2.5, and Qwen3 are supported.

Example:

```python
>>> from trl.chat_template_utils import get_training_chat_template
>>> from transformers import AutoTokenizer

>>> tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-0.6B")
>>> messages1 = [
...     {"role": "user", "content": "What is 2 * 3?"},
...     {
...         "role": "assistant",
...         "content": "",
...         "tool_calls": [{"type": "function", "function": {"name": "multiply", "arguments": {"a": 2, "b": 3}}}],
...     },
... ]
>>> messages2 = messages1 + [
...     {"role": "tool", "name": "multiply", "content": "6"},
... ]
>>> tokenizer.apply_chat_template(messages1, tokenize=False)
'user\nWhat is 2 * 3?\nassistant\n\n\n\n\n\n{"name": "multiply", "arguments": {"a": 2, "b": 3}}\n\n'

>>> tokenizer.apply_chat_template(messages2, tokenize=False, add_generation_prompt=True)
'user\nWhat is 2 * 3?\nassistant\n\n{"name": "multiply", "arguments": {"a": 2, "b": 3}}\n\nuser\n\n6\n\nassistant\n'

>>> #                                                        ^ think tags missing
>>> chat_template = get_training_chat_template(tokenizer)
>>> tokenizer.apply_chat_template(messages1, tokenize=False, chat_template=chat_template)
'user\nWhat is 2 * 3?\nassistant\n\n\n\n\n\n{"name": "multiply", "arguments": {"a": 2, "b": 3}}\n\n'

>>> tokenizer.apply_chat_template(
...     messages2, tokenize=False, add_generation_prompt=True, chat_template=chat_template
... )
'user\nWhat is 2 * 3?\nassistant\n\n\n\n\n\n{"name": "multiply", "arguments": {"a": 2, "b": 3}}\n\nuser\n\n6\n\nassistant\n'
```

**Parameters:**

tokenizer (`PreTrainedTokenizer`) : Tokenizer instance to check.

**Returns:**

``str` or `None``

Training-compatible chat template, or `None` if no patching is needed.