Add full tool calling support to chat template

This adds full tool calling support to the chat template, producing pretty much identical results as the the prompt processing code in `tokenization_minicpm.py`. Fully supports generating `tool_call` (requires `arguments` to be a dict (as supported in recent `transformers`) and not a string) and `thought` history.

Also sets `add_bos_token` to false to prevent `<s>` from being added to the prompt when not using the chat template, as it clearly should not be.

tokenizer_config.json

@@ -2,7 +2,7 @@
   "auto_map": {
     "AutoTokenizer": ["tokenization_minicpm.MiniCPMTokenizer", null]
   },
-  "add_bos_token": true,
+  "add_bos_token": false,
   "add_eos_token": false,
   "added_tokens_decoder": {
     "0": {
@@ -115,5 +115,5 @@
   "tokenizer_class": "MiniCPMTokenizer",
   "unk_token": "<unk>",
   "use_default_system_prompt": false,
-  "chat_template": "{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}"
+  "chat_template": "{%- macro json_to_python_type(param_name, json_spec) %}\n{%- set basic_type_map = {\n  'string': 'str',\n  'number': 'float',\n  'integer': 'int',\n  'boolean': 'bool'\n} %}\n\n{%- if json_spec.enum %}\n  {{- param_name|title }}\n{%- elif basic_type_map[json_spec.type] is defined %}\n  {{- basic_type_map[json_spec.type] }}\n{%- elif json_spec.type == 'array' %}\n  {{- 'List[' +  json_to_python_type(param_name, json_spec.items) + ']' }}\n{%- elif json_spec.type == 'object' %}\n  {{- 'Dict[str, ' + json_to_python_type(param_name, json_spec.additionalProperties) + ']' }}\n{%- elif json_spec.type is iterable %}\n  {{- 'Union[' }}\n  {%- for t in json_spec.type %}\n    {{- json_to_python_type(param_name, {'type': t}) }}\n    {{- ', ' if not loop.last }}\n  {%- endfor %}\n    {{- ']' }}\n{%- else %}\n  {{- 'Any' }}\n{%- endif %}\n{%- endmacro %}\n\n{%- macro tool_parser(tools) %}\n{%- for tool in tools %}\n  {%- if tool.function is defined %}\n    {%- set tool = tool.function %}\n  {%- endif %}\n  {%- for param_name, param_fields in tool.parameters.properties|items %}\n    {%- if param_fields.enum %}\n      {{- '\\n\\nclass ' + param_name|title + '(Enum):\\n' }}\n      {%- for enum_option in param_fields.enum %}\n        {{- '    enum_' + loop.index0|string + ' = ' + enum_option|tojson + '\\n' }}\n      {%- endfor %}\n    {%- endif %}\n  {%- endfor %}\n  {{- '\\n\\ndef ' + tool.name + '(' }}\n  {%- for param_name, param_fields in tool.parameters.properties|items %}\n    {{- ', ' if loop.index0 != 0 }}\n    {{- param_name }} \n    {{- ' = None' if param_name not in tool.parameters.required }}\n  {%- endfor %}\n  {{- '):\\n    \"\"\"' }}\n  {{- tool.description }}\n  {%- if tool.parameters.properties|length != 0 %}\n    {{- '\\n\\n    Args:\\n        ' }}\n    {%- for param_name, param_fields in tool.parameters.properties|items %}\n      {{- '\\n        ' if loop.index0 != 0 }}\n      {{- param_name + ': ' }}\n      {%- if param_name not in tool.parameters.required %}\n        {{- 'Optional[' + json_to_python_type(param_name, param_fields) + ']' }}\n      {%- else %}\n        {{- json_to_python_type(param_name, param_fields) }}\n      {%- endif %}\n      {%- if not param_fields.description %}\n        {{- ' = None' if param_name not in tool.parameters.required }}\n      {%- else %}\n        {{- ' = Field(..., ' if param_name in tool.parameters.required else ' = Field(None, ' }}\n        {{- 'description=' + param_fields.description|tojson + ')' }}\n      {%- endif %}\n    {%- endfor %}\n  {%- endif %}\n  {{- '\\n    \"\"\"\\n' }}\n{%- endfor %}\n{%- endmacro %}\n\n{%- if messages[0]['role'] == 'system' %}\n  {%- set loop_messages = messages[1:] %}\n  {%- set system_message = messages[0]['content'] %}\n{%- else %}\n  {%- set loop_messages = messages %}\n  {%- set system_message = '' %}\n{%- endif %}\n{{- '<|im_start|>system\\n' + system_message if system_message or tools }}\n{%- if tools %}\n  {{- '\\n# Functions\\nHere is a list of functions that you can invoke:\\n```python\\nfrom enum import Enum\\nfrom typing import List, Dict, Optional\\nfrom pydantic import BaseModel, Field\\n\\n' }}\n    {{- tool_parser(tools) }}\n  {{- \"\\n```\\n\\n# Function Call Rule and Output Format\\n- If the user's question can be answered without calling any function, please answer the user's question directly. In this situation, you should return your thought and answer the user's question directly.\\n- If the user cannot be answered without calling any function, and the user does not provide enough information to call functions, please ask the user for more information. In this situation, you should return your thought and ask the user for more information.\\n- If the user's question cannot be answered without calling any function, and the user has provided enough information to call functions to solve it, you should call the functions. In this situation, the assistant should return your thought and call the functions.\\n- Use default parameters unless the user has specified otherwise.\\n- You should answer in the following format:\\n\\n<|thought_start|>\\n{explain why the user's question can be answered without calling a function or why you should ask the user for more information or why you should call one or more functions and your plan to solve the user's question.}\\n<|thought_end|>\\n<|tool_call_start|>\\n```python\\nfunc1(params_name=params_value, params_name2=params_value2...)\\nfunc2(params)\\n```\\n<|tool_call_end|>\\n{answer the user's question directly or ask the user for more information}\" }}\n{%- endif %}\n{{- '<|im_end|>\\n' }}\n{%- for message in loop_messages %}\n  {%- set content = message.content %}\n  {%- if message.role == 'assistant' and message.tool_calls %}\n    {{- '<|im_start|>' + message.role + '\\n' }}\n    {{- '<|thought_start|>\\n' + message.thought + '\\n<|thought_end|>\\n' if message.thought }}\n    {{- '<|tool_call_start|>\\n```python\\n' }}\n    {%- for tool_call in message.tool_calls %}\n      {%- if tool_call.function is defined %}\n        {%- set tool_call = tool_call.function %}\n      {%- endif %}\n      {{- tool_call.name + '(' }}\n      {%- if tool_call.arguments is defined and tool_call.arguments|length > 0 %}\n        {%- for param_name, param_value in tool_call.arguments|items %}\n          {{- param_name + '=' + param_value|tojson }}\n          {{- ',' if not loop.last }}\n        {%- endfor %}\n      {%- endif %}\n      {{- ')\\n' }}\n    {%- endfor %}\n    {{- '```\\n<|tool_call_end|>\\n' }}\n    {{- content if content and not content.startswith('<|tool_call_start|>') }}\n    {{- '<|im_end|>\\n' }}\n  {%- elif message.role == 'assistant' and message.thought %}\n    {{- '<|im_start|>' + message.role + '\\n' + '<|thought_start|>\\n' + message.thought + '\\n<|thought_end|>\\n' + content + '<|im_end|>\\n' }}\n  {%- else %}\n    {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>\\n' }}\n  {%- endif %}\n{%- endfor %}\n\n{%- if add_generation_prompt %}\n  {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n"
 }