lmflow.utils.conversation_template

Submodules

• lmflow.utils.conversation_template.base
• lmflow.utils.conversation_template.chatglm
• lmflow.utils.conversation_template.chatml
• lmflow.utils.conversation_template.deepseek
• lmflow.utils.conversation_template.gemma
• lmflow.utils.conversation_template.hymba
• lmflow.utils.conversation_template.internlm
• lmflow.utils.conversation_template.llama
• lmflow.utils.conversation_template.phi
• lmflow.utils.conversation_template.qwen
• lmflow.utils.conversation_template.yi
• lmflow.utils.conversation_template.zephyr

Attributes

EMPTY_TEMPLATE
EMPTY_NO_SPECIAL_TOKENS_TEMPLATE
CHATGLM3_TEMPLATE
CHATML_TEMPLATE
DEEPSEEK_V2_TEMPLATE
DEEPSEEK_V3_TEMPLATE
DEEPSEEK_R1_TEMPLATE
DEEPSEEK_R1_DISTILL_TEMPLATE
GEMMA_TEMPLATE
HYMBA_TEMPLATE
INTERNLM2_TEMPLATE
LLAMA2_TEMPLATE
LLAMA3_TEMPLATE
LLAMA3_TEMPLATE_FOR_TOOL
PHI3_TEMPLATE
QWEN2_TEMPLATE
QWEN2_TEMPLATE_FOR_TOOL
QWEN2_5_TEMPLATE
QWEN2_5_1M_TEMPLATE
QWEN2_5_MATH_TEMPLATE
QWEN_QWQ_TEMPLATE
YI1_5_TEMPLATE
ZEPHYR_TEMPLATE
logger
PRESET_TEMPLATES
JINJA_TEMPLATES

Classes

ConversationTemplate
ConversationTemplateForTool

Functions

is_package_version_at_least(package_name, min_version)

Package Contents

lmflow.utils.conversation_template.is_package_version_at_least(package_name, min_version)

lmflow.utils.conversation_template.EMPTY_TEMPLATE

lmflow.utils.conversation_template.EMPTY_NO_SPECIAL_TOKENS_TEMPLATE

class lmflow.utils.conversation_template.ConversationTemplate

user_formatter: Formatter

assistant_formatter: Formatter

function_formatter: Formatter | None = None

observation_formatter: Formatter | None = None

system_formatter: Formatter | None = None

force_system: bool = False

tools_formatter: Formatter | None = None

separator: TemplateComponent | None = None

remove_last_sep: bool = False

special_starter: TemplateComponent | None = None

special_stopper: TemplateComponent | None = None

template_name: str | None = None

system_default: str | None = None

__post_init__()

encode_conversation(tokenizer: transformers.PreTrainedTokenizer, messages: List[Dict[str, str]], system: str | None = None, tools: List[str] | None = None, **kwargs) → Sequence[Tuple[List[int], List[int]]]

Messages here should be guaranteed to be in pairs, with the first message being the user message and the second message being the system message. Data example: ```json {

“conversation_id”: 2, “system”: “sysinfo1”, “tools”: [“tool_1_desc”], “messages”: [

{

“role”: “user”, “content”: “hi”

}, {

“role”: “assistant”, “content”: “Hello!”

}

]

}

_encode(tokenizer: transformers.PreTrainedTokenizer, messages: List[Dict[str, str]], system: str | None = None, tools: str | None = None, **kwargs) → Sequence[Tuple[List[int], List[int]]]

_encode_template(template: List[TemplateComponent], tokenizer: transformers.PreTrainedTokenizer, **kwargs) → List[int]

Encode template components into token ids.

Parameters:

templateList[TemplateComponent]

Formatted template components.

tokenizerPreTrainedTokenizer

Tokenizer to convert tokens into token ids.

Returns:

List[int]

Encoded token ids.

post_process_pairs(encoded_pairs, tokenizer)

remove_last_separator(encoded_pairs: Sequence[Tuple[List[int], List[int]]], tokenizer: transformers.PreTrainedTokenizer) → Sequence[Tuple[List[int], List[int]]]

add_special_starter(encoded_pairs: Sequence[Tuple[List[int], List[int]]], tokenizer: transformers.PreTrainedTokenizer) → Sequence[Tuple[List[int], List[int]]]

add_special_stopper(encoded_pairs: Sequence[Tuple[List[int], List[int]]], tokenizer: transformers.PreTrainedTokenizer) → Sequence[Tuple[List[int], List[int]]]

_ensure_id_list(obj: int | List[int]) → List[int]

Make sure the object is a list of integers. Useful for handling token ids.

class lmflow.utils.conversation_template.ConversationTemplateForTool

Bases: ConversationTemplate

encode_conversation(tokenizer: transformers.PreTrainedTokenizer, messages: List[Dict[str, str]], system: str | None = None, tools: List[str] | None = None, **kwargs) → Sequence[Tuple[List[int], List[int]]]

Messages here should be guaranteed to be in pairs, with the first message being the user message and the second message being the system message. Data example: ```json {

“conversation_id”: 2, “system”: “sysinfo1”, “tools”: [“tool_1_desc”], “messages”: [

{

“role”: “user”, “content”: “hi”

}, {

“role”: “assistant”, “content”: “Hello!”

}

]

}

_encode(tokenizer: transformers.PreTrainedTokenizer, messages: List[Dict[str, str]], system: str | None = None, tools: str | None = None, **kwargs) → Sequence[Tuple[List[int], List[int]]]

_encode_template(template: List[TemplateComponent], tokenizer: transformers.PreTrainedTokenizer, **kwargs) → List[int]

Encode template components into token ids.

Parameters:

templateList[TemplateComponent]

Formatted template components.

tokenizerPreTrainedTokenizer

Tokenizer to convert tokens into token ids.

Returns:

List[int]

Encoded token ids.

_handle_tools(tools: List[str] | None) → str

lmflow.utils.conversation_template.CHATGLM3_TEMPLATE

lmflow.utils.conversation_template.CHATML_TEMPLATE

lmflow.utils.conversation_template.DEEPSEEK_V2_TEMPLATE

lmflow.utils.conversation_template.DEEPSEEK_V3_TEMPLATE = Multiline-String

Show Value

"""{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% set ns = namespace(is_first=false, is_tool=false, is_output_first=true, system_prompt='', is_first_sp=true) %}{%- for message in messages %}{%- if message['role'] == 'system' %}{%- if ns.is_first_sp %}{% set ns.system_prompt = ns.system_prompt + message['content'] %}{% set ns.is_first_sp = false %}{%- else %}{% set ns.system_prompt = ns.system_prompt + '

' + message['content'] %}{%- endif %}{%- endif %}{%- endfor %}{{bos_token}}{{ns.system_prompt}}{%- for message in messages %}{%- if message['role'] == 'user' %}{%- set ns.is_tool = false -%}{{'<｜User｜>' + message['content']}}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is none %}{%- set ns.is_tool = false -%}{%- for tool in message['tool_calls']%}{%- if not ns.is_first %}{{'<｜Assistant｜>'}}{% generation %}{{'<｜tool▁calls▁begin｜><｜tool▁call▁begin｜>' + tool['type'] + '<｜tool▁sep｜>' + tool['function']['name'] + '
' + '```json' + '
' + tool['function']['arguments'] + '
' + '```' + '<｜tool▁call▁end｜>'}}{% endgeneration %}{%- set ns.is_first = true -%}{%- else %}{% generation %}{{'
' + '<｜tool▁call▁begin｜>' + tool['type'] + '<｜tool▁sep｜>' + tool['function']['name'] + '
' + '```json' + '
' + tool['function']['arguments'] + '
' + '```' + '<｜tool▁call▁end｜>'}}{{'<｜tool▁calls▁end｜><｜end▁of▁sentence｜>'}}{% endgeneration %}{%- endif %}{%- endfor %}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is not none %}{%- if ns.is_tool %}{{'<｜tool▁outputs▁end｜>'}}{% generation %}{{ message['content'] + '<｜end▁of▁sentence｜>'}}{%- set ns.is_tool = false -%}{% endgeneration %}{%- else %}{{'<｜Assistant｜>'}}{% generation %}{{ message['content'] + '<｜end▁of▁sentence｜>'}}{% endgeneration %}{%- endif %}{%- endif %}{%- if message['role'] == 'tool' %}{%- set ns.is_tool = true -%}{%- if ns.is_output_first %}{{'<｜tool▁outputs▁begin｜><｜tool▁output▁begin｜>' + message['content'] + '<｜tool▁output▁end｜>'}}{%- set ns.is_output_first = false %}{%- else %}{{'
<｜tool▁output▁begin｜>' + message['content'] + '<｜tool▁output▁end｜>'}}{%- endif %}{%- endif %}{%- endfor -%}{% if ns.is_tool %}{{'<｜tool▁outputs▁end｜>'}}{% endif %}{% if add_generation_prompt and not ns.is_tool %}{{'<｜Assistant｜>'}}{% endif %}"""

lmflow.utils.conversation_template.DEEPSEEK_R1_TEMPLATE = "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif...

lmflow.utils.conversation_template.DEEPSEEK_R1_DISTILL_TEMPLATE = "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif...

lmflow.utils.conversation_template.GEMMA_TEMPLATE

lmflow.utils.conversation_template.HYMBA_TEMPLATE

lmflow.utils.conversation_template.INTERNLM2_TEMPLATE

lmflow.utils.conversation_template.LLAMA2_TEMPLATE

lmflow.utils.conversation_template.LLAMA3_TEMPLATE

lmflow.utils.conversation_template.LLAMA3_TEMPLATE_FOR_TOOL

lmflow.utils.conversation_template.PHI3_TEMPLATE

lmflow.utils.conversation_template.QWEN2_TEMPLATE

lmflow.utils.conversation_template.QWEN2_TEMPLATE_FOR_TOOL

lmflow.utils.conversation_template.QWEN2_5_TEMPLATE = '{%- if tools %}{{- \'<|im_start|>system\\n\' }}{%- if messages[0][\'role\'] == \'system\' %}{{-...

lmflow.utils.conversation_template.QWEN2_5_1M_TEMPLATE = '{%- if tools %}{{- \'<|im_start|>system\\n\' }}{%- if messages[0][\'role\'] == \'system\' %}{{-...

lmflow.utils.conversation_template.QWEN2_5_MATH_TEMPLATE = '{%- if tools %}{{- \'<|im_start|>system\\n\' }}{%- if messages[0][\'role\'] == \'system\' %}{{-...

lmflow.utils.conversation_template.QWEN_QWQ_TEMPLATE = '{%- if tools %}{{- \'<|im_start|>system\\n\' }}{%- if messages[0][\'role\'] == \'system\' %}{{-...

lmflow.utils.conversation_template.YI1_5_TEMPLATE

lmflow.utils.conversation_template.ZEPHYR_TEMPLATE

lmflow.utils.conversation_template.logger

lmflow.utils.conversation_template.PRESET_TEMPLATES

lmflow.utils.conversation_template.JINJA_TEMPLATES