Rethinking Context Management in Language Models: Beyond Conversational Turns
By
nemath
4mo ago· 10 min readenInsight
100/100
Golden Brown
Bagelometer↗
The kind of bagel that ruins lesser bagels for you.
Score100TypeanalysisSentimentneutral
Summary
The article discusses approaches to providing continuous context to language models, challenging the conventional conversational turn-based paradigm. It argues that every request to a model already includes the full context history, and users can arbitrarily edit or manipulate this history. The author suggests thinking beyond conversational turns and considering LLM interactions as one-shot prompts where context can be completely controlled and reconstructed by the user.
Key quotes
· 4 pulledEvery time you send a request to a model you're already providing all of the context history along with it.
To edit the context, just send a different context history. You can send whatever you want as history, it's entirely up to you and entirely arbitrary.
We only think in conversational turns because that's what we've expected a conversation to 'look like'. But that's just a very deeply ingrained convention.
Forget that there is such a thing as 'turns' in a LLM convo for now, imagine that it's all 'one-shot'.
Every time you send a request to a model you're already providing all of the context history along with it. To edit the context, just send a different context history. You can send whatever you want as history, it's entirely up to you and entirely arbitra
