Inspired by and featuring the Reflection Tuning technique pioneered by Matt Shumer (possibly earlier innovated by the team at Anthropic.)

'''

During sampling, the model will start by outputting reasoning inside and tags, and then once it is satisfied with its reasoning, it will output the final answer inside

and

tags. Each of these tags are special tokens, trained into the model.

This enables the model to separate its internal thoughts and reasoning from its final answer, improving the experience for the user.

Inside the section, the model may output one or more tags, which signals the model has caught an error in its reasoning and will attempt to correct it before providing a final answer.

System Prompt: The system prompt used for training this model is:

You are a world-class AI system, capable of complex reasoning and reflection. Reason through the query inside tags, and then provide your final response inside

tags. If you detect that you made a mistake in your reasoning at any point, correct yourself inside tags.

We recommend using this exact system prompt to get the best results from Reflection Llama-3.1 70B. You may also want to experiment combining this system prompt with your own custom instructions to customize the behavior of the model.

Chat Format: As mentioned above, the model uses the standard Llama 3.1 chat format. Here’s an example:

<|begin_of_text|><|start_header_id|>system<|end_header_id|>

You are a world-class AI system, capable of complex reasoning and reflection. Reason through the query inside tags, and then provide your final response inside

tags. If you detect that you made a mistake in your reasoning at any point, correct yourself inside tags.<|eot_id|><|start_header_id|>user<|end_header_id|>

what is 2+2?<|eot_id|><|start_header_id|>assistant<|end_header_id|>

'''

This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.