x

LESSWRONG
LW

Kilian Merkelbach

Subscribe

Message

4

1

7mo

Prompt Framing Changes LLM Performance (and Safety)

Let's investigate the effects of prompt framing on LLM (large language model) performance and safety. For instance, does a model's performance increase or decrease if it receives a call to urgency before being given a task? Is it an effective strategy to instruct the model to take a deep breath...

Oct 2, 20255

Kilian Merkelbach

Subscribe

Message

4

1

7mo

Prompt Framing Changes LLM Performance (and Safety)

Let's investigate the effects of prompt framing on LLM (large language model) performance and safety. For instance, does a model's performance increase or decrease if it receives a call to urgency before being given a task? Is it an effective strategy to instruct the model to take a deep breath...

Oct 2, 20255

Prompt Framing Changes LLM Performance (and Safety)

Kilian Merkelbach

4mo

Let's investigate the effects of prompt framing on LLM (large language model) performance and safety.
For instance, does a model's performance increase or decrease if it receives a call to urgency before being given a task? Is it an effective strategy to instruct the model to take a deep breath before providing its answer?

Summary

We consider the effect of simple prompt framing on the performance and safety of LLMs. The task of coding profits most strongly from framing. The most reliable form of framing is to instruct the model to take a deep breath before giving its reply, which works well even for recall or logic-based tasks. Prompt framing can also influence a model's... (read 2053 more words →)

5