chat_template.jinja
| 1 | [gMASK]<sop> |
| 2 | {%- if tools -%} |
| 3 | <|system|> |
| 4 | # Tools |
| 5 | |
| 6 | You may call one or more functions to assist with the user query. |
| 7 | |
| 8 | You are provided with function signatures within <tools></tools> XML tags: |
| 9 | <tools> |
| 10 | {% for tool in tools %} |
| 11 | {{ tool | tojson(ensure_ascii=False) }} |
| 12 | {% endfor %} |
| 13 | </tools> |
| 14 | |
| 15 | For each function call, output the function name and arguments within the following XML format: |
| 16 | <tool_call>{function-name} |
| 17 | <arg_key>{arg-key-1}</arg_key> |
| 18 | <arg_value>{arg-value-1}</arg_value> |
| 19 | <arg_key>{arg-key-2}</arg_key> |
| 20 | <arg_value>{arg-value-2}</arg_value> |
| 21 | ... |
| 22 | </tool_call>{%- endif -%} |
| 23 | {%- macro visible_text(content) -%} |
| 24 | {%- if content is string -%} |
| 25 | {{- content }} |
| 26 | {%- elif content is iterable and content is not mapping -%} |
| 27 | {%- for item in content -%} |
| 28 | {%- if item is mapping and item.type == 'text' -%} |
| 29 | {{- item.text }} |
| 30 | {%- elif item is mapping and (item.type == 'image' or 'image' in item) -%} |
| 31 | <|begin_of_image|><|image|><|end_of_image|> |
| 32 | {%- elif item is mapping and (item.type == 'video' or 'video' in item) -%} |
| 33 | <|begin_of_video|><|video|><|end_of_video|> |
| 34 | {%- elif item is string -%} |
| 35 | {{- item }} |
| 36 | {%- endif -%} |
| 37 | {%- endfor -%} |
| 38 | {%- else -%} |
| 39 | {{- content }} |
| 40 | {%- endif -%} |
| 41 | {%- endmacro -%} |
| 42 | {%- set ns = namespace(last_user_index=-1) %} |
| 43 | {%- for m in messages %} |
| 44 | {%- if m.role == 'user' %} |
| 45 | {% set ns.last_user_index = loop.index0 -%} |
| 46 | {%- endif %} |
| 47 | {%- endfor %} |
| 48 | {% for m in messages %} |
| 49 | {%- if m.role == 'user' -%}<|user|> |
| 50 | {% if m.content is string %} |
| 51 | {{ m.content }} |
| 52 | {%- else %} |
| 53 | {%- for item in m.content %} |
| 54 | {% if item.type == 'video' or 'video' in item %} |
| 55 | <|begin_of_video|><|video|><|end_of_video|>{% elif item.type == 'image' or 'image' in item %} |
| 56 | <|begin_of_image|><|image|><|end_of_image|>{% elif item.type == 'text' %} |
| 57 | {{ item.text }} |
| 58 | {%- endif %} |
| 59 | {%- endfor %} |
| 60 | {%- endif %} |
| 61 | {{- '/nothink' if (enable_thinking is defined and not enable_thinking and not visible_text(m.content).endswith("/nothink")) else '' -}} |
| 62 | {%- elif m.role == 'assistant' -%} |
| 63 | <|assistant|> |
| 64 | {%- set reasoning_content = '' %} |
| 65 | {%- set content = visible_text(m.content) %} |
| 66 | {%- if m.reasoning_content is string %} |
| 67 | {%- set reasoning_content = m.reasoning_content %} |
| 68 | {%- else %} |
| 69 | {%- if '</think>' in content %} |
| 70 | {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %} |
| 71 | {%- set content = content.split('</think>')[-1].lstrip('\n') %} |
| 72 | {%- endif %} |
| 73 | {%- endif %} |
| 74 | {%- if loop.index0 > ns.last_user_index and reasoning_content -%} |
| 75 | {{ '\n<think>' + reasoning_content.strip() + '</think>'}} |
| 76 | {%- else -%} |
| 77 | {{ '\n<think></think>' }} |
| 78 | {%- endif -%} |
| 79 | {%- if content.strip() -%} |
| 80 | {{ '\n' + content.strip() }} |
| 81 | {%- endif -%} |
| 82 | {% if m.tool_calls %} |
| 83 | {% for tc in m.tool_calls %} |
| 84 | {%- if tc.function %} |
| 85 | {%- set tc = tc.function %} |
| 86 | {%- endif %} |
| 87 | {{ '\n<tool_call>' + tc.name }} |
| 88 | {% set _args = tc.arguments %} |
| 89 | {% for k, v in _args.items() %} |
| 90 | <arg_key>{{ k }}</arg_key> |
| 91 | <arg_value>{{ v | tojson(ensure_ascii=False) if v is not string else v }}</arg_value> |
| 92 | {% endfor %} |
| 93 | </tool_call>{% endfor %} |
| 94 | {% endif %} |
| 95 | {%- elif m.role == 'tool' -%} |
| 96 | {%- if m.content is string -%} |
| 97 | {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %} |
| 98 | {{- '<|observation|>' }} |
| 99 | {%- endif %} |
| 100 | {{- '\n<tool_response>\n' }} |
| 101 | {{- m.content }} |
| 102 | {{- '\n</tool_response>' }} |
| 103 | {% elif m.content is iterable and m.content is not mapping %} |
| 104 | {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %} |
| 105 | {{- '<|observation|>' }} |
| 106 | {%- endif %} |
| 107 | {{- '\n<tool_response>\n' }} |
| 108 | {%- for tr in m.content -%} |
| 109 | {%- if tr is mapping and tr.type is defined -%} |
| 110 | {%- set t = tr.type | lower -%} |
| 111 | {%- if t == 'text' and tr.text is defined -%} |
| 112 | {{ tr.text }} |
| 113 | {%- elif t in ['image', 'image_url'] -%} |
| 114 | <|begin_of_image|><|image|><|end_of_image|> |
| 115 | {%- elif t in ['video', 'video_url'] -%} |
| 116 | <|begin_of_video|><|video|><|end_of_video|> |
| 117 | {%- else -%} |
| 118 | {{ tr | tojson(ensure_ascii=False) }} |
| 119 | {%- endif -%} |
| 120 | {%- else -%} |
| 121 | {{ tr.output if tr.output is defined else tr }} |
| 122 | {%- endif -%} |
| 123 | {%- endfor -%} |
| 124 | {{- '\n</tool_response>' }} |
| 125 | {%- else -%} |
| 126 | <|observation|>{% for tr in m.content %} |
| 127 | |
| 128 | <tool_response> |
| 129 | {{ tr.output if tr.output is defined else tr }} |
| 130 | </tool_response>{% endfor -%} |
| 131 | {% endif -%} |
| 132 | {%- elif m.role == 'system' -%} |
| 133 | <|system|> |
| 134 | {{ visible_text(m.content) }} |
| 135 | {%- endif -%} |
| 136 | {%- endfor -%} |
| 137 | {%- if add_generation_prompt -%} |
| 138 | <|assistant|> |
| 139 | {{'<think></think>\n' if (enable_thinking is defined and not enable_thinking) else ''}} |
| 140 | {%- endif -%} |
| 141 | |