phi-4-reasoning chat template forces system prompt?
#6
by
RyanMetcalfeInt8
- opened
Hello, in tokenizer_config.json, the chat template is defined as:
"chat_template": "<|im_start|>system<|im_sep|>You are Phi, a language model trained by Microsoft to help users. Your role as an assistant involves thoroughly exploring questions through a systematic thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution using the specified format: <think> {Thought section} </think> {Solution section}. In the Thought section, detail your reasoning process in steps. Each step should include detailed considerations such as analysing questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The Solution section should be logical, accurate, and concise and detail necessary steps needed to reach the conclusion. Now, try to solve the following question through the above guidelines:<|im_end|>{% for message in messages %}{% if (message['role'] == 'user') %}{{'<|im_start|>user<|im_sep|>' + message['content'] + '<|im_end|>'}}{% elif (message['role'] == 'assistant') %}{{'<|im_start|>assistant<|im_sep|>'}}{% generation %}{{message['content'] + '<|im_end|>'}}{% endgeneration %}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant<|im_sep|>' }}{% endif %}",
Is this intended? It seems that there is no way to not add the system prompt. For example, it will get added even when system role is not defined:
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True
)
This is problematic when apply_chat_template is used to apply chat template for 2nd prompt & onward in a chat sequence, where pipeline maintains history (in kvcache).