Llama 3.1 models in prompting mode have a slightly different chat template than Llama 3.0. Specifically, they need to carry the Cutting Knowledge Date and the Today Date information in the system prompt. This PR accounts for that difference in the Llama prompting handler (llama.py).
Here is the difference in how we should format the system prompt:
formatted_prompt += "<|start_header_id|>system<|end_header_id|>\n\n"
formatted_prompt += "Cutting Knowledge Date: December 2023\n" # Llama 3.0 doesn't have this line
formatted_prompt += "Today Date: 26 Jul 2024\n\n" # Llama 3.0 doesn't have this line
formatted_prompt += system_message + "<|eot_id|>"
Note: Although Llama 3.2 has the same chat template as Llama 3.1, their model card doesn't contain such additional information, so we adhere to that, and thus Llama 3.2 still have the same processing logic with Llama 3.0.
For reference, below is the Llama 3.0 chat template.
"bos_token": "<|begin_of_text|>",
"chat_template": "{% set loop_messages = messages %}{% for message in loop_messages %}{% set content = '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' %}{% if loop.index0 == 0 %}{% set content = bos_token + content %}{% endif %}{{ content }}{% endfor %}{% if add_generation_prompt %}{{ '<|start_header_id|>assistant<|end_header_id|>\n\n' }}{% endif %}"
And this is the Llama 3.1 chat template.
"bos_token": "<|begin_of_text|>",
"chat_template":
{{- bos_token }}
{%- if custom_tools is defined %}
{%- set tools = custom_tools %}
{%- endif %}
{%- if not tools_in_user_message is defined %}
{%- set tools_in_user_message = true %}
{%- endif %}
{%- if not date_string is defined %}
{%- set date_string = "26 Jul 2024" %}
{%- endif %}
{%- if not tools is defined %}
{%- set tools = none %}
{%- endif %}
{#- This block extracts the system message, so we can slot it into the right place. #}
{%- if messages[0]['role'] == 'system' %}
{%- set system_message = messages[0]['content']|trim %}
{%- set messages = messages[1:] %}
{%- else %}
{%- set system_message = "" %}
{%- endif %}
{#- System message + builtin tools #}
{{- "<|start_header_id|>system<|end_header_id|>\n\n" }}
{%- if builtin_tools is defined or tools is not none %}
{{- "Environment: ipython\n" }}
{%- endif %}
{%- if builtin_tools is defined %}
{{- "Tools: " + builtin_tools | reject('equalto', 'code_interpreter') | join(", ") + "\n\n"}}
{%- endif %}
{{- "Cutting Knowledge Date: December 2023\n" }}
{{- "Today Date: " + date_string + "\n\n" }}
{%- if tools is not none and not tools_in_user_message %}
{{- "You have access to the following functions. To call a function, please respond with JSON for a function call." }}
{{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}
{{- "Do not use variables.\n\n" }}
{%- for t in tools %}
{{- t | tojson(indent=4) }}
{{- "\n\n" }}
{%- endfor %}
{%- endif %}
{{- system_message }}
{{- "<|eot_id|>" }}
{#- Custom tools are passed in a user message with some extra guidance #}
{%- if tools_in_user_message and not tools is none %}
{#- Extract the first user message so we can plug it in here #}
{%- if messages | length != 0 %}
{%- set first_user_message = messages[0]['content']|trim %}
{%- set messages = messages[1:] %}
{%- else %}
{{- raise_exception("Cannot put tools in the first user message when there's no first user message!") }}
{%- endif %}
{{- '<|start_header_id|>user<|end_header_id|>\n\n' -}}
{{- "Given the following functions, please respond with a JSON for a function call " }}
{{- "with its proper arguments that best answers the given prompt.\n\n" }}
{{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}
{{- "Do not use variables.\n\n" }}
{%- for t in tools %}
{{- t | tojson(indent=4) }}
{{- "\n\n" }}
{%- endfor %}
{{- first_user_message + "<|eot_id|>"}}
{%- endif %}
{%- for message in messages %}
{%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}
{{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' }}
{%- elif 'tool_calls' in message %}
{%- if not message.tool_calls|length == 1 %}
{{- raise_exception("This model only supports single tool-calls at once!") }}
{%- endif %}
{%- set tool_call = message.tool_calls[0].function %}
{%- if builtin_tools is defined and tool_call.name in builtin_tools %}
{{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}
{{- "<|python_tag|>" + tool_call.name + ".call(" }}
{%- for arg_name, arg_val in tool_call.arguments | items %}
{{- arg_name + '="' + arg_val + '"' }}
{%- if not loop.last %}
{{- ", " }}
{%- endif %}
{%- endfor %}
{{- ")" }}
{%- else %}
{{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}
{{- '{"name": "' + tool_call.name + '", ' }}
{{- '"parameters": ' }}
{{- tool_call.arguments | tojson }}
{{- "}" }}
{%- endif %}
{%- if builtin_tools is defined %}
{#- This means we're in ipython mode #}
{{- "<|eom_id|>" }}
{#- This means we're in ipython mode #}
{{- "<|eom_id|>" }}
{{- "<|eom_id|>" }}
{%- else %}
{{- "<|eot_id|>" }}
{%- endif %}
{%- elif message.role == "tool" or message.role == "ipython" %}
{{- "<|start_header_id|>ipython<|end_header_id|>\n\n" }}
{%- if message.content is mapping or message.content is iterable %}
{{- message.content | tojson }}
{%- else %}
{{- message.content }}
{%- endif %}
{{- "<|eot_id|>" }}
{%- endif %}
{%- endfor %}
{%- if add_generation_prompt %}
{{- '<|start_header_id|>assistant<|end_header_id|>\n\n' }}
{%- endif %}
Llama 3.1 models in prompting mode have a slightly different chat template than Llama 3.0. Specifically, they need to carry the
Cutting Knowledge Dateand theToday Dateinformation in the system prompt. This PR accounts for that difference in the Llama prompting handler (llama.py).Here is the difference in how we should format the system prompt:
Note: Although Llama 3.2 has the same chat template as Llama 3.1, their model card doesn't contain such additional information, so we adhere to that, and thus Llama 3.2 still have the same processing logic with Llama 3.0.
For reference, below is the Llama 3.0 chat template.
And this is the Llama 3.1 chat template.