Paper frames instruction-tuned models as an instructional-conversational genre
Original title: The Language of Instruction: Genre and Generative AI
The argument ties the genre to instruction fine-tuning, system prompts, special tokens and role-based turn-taking.
An arXiv article argues that instruction-fine-tuned language models implement an instructional-conversational genre. It describes that genre as an amalgamation of earlier human–model conversation, recursively processed by large neural networks. The paper says the genre appears in applications through instruction fine-tuning, a system-prompt pretext, and markup that uses special tokens and role assignment in turn-taking.
- The proposed genre mixes prior human–model conversational language.
- Instantiation includes a system prompt and formalized role tokens.
- The piece is a genre argument rather than a new model or benchmark.
From the original article
1 min read at arxiv.orgThis article argues that instruction-fine-tuned language models should be understood as implementing a genre, what is proposed as the instructional-conversational genre. This genre is an amalgamation of discursive language from prior conversations between generative models and humans recursively run through these massive neural networks.
By James E. Dobson · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.