Stop Anthropomorphizing Intermediate Tokens as Reasoning/Thinking Traces
Why CEREBRO kept it
Token interpretation analysis, LLM mechanics deep-dive.
The text below is an automated extraction of the article at https://arxiv.org/abs/2504.09762, stored verbatim in the public cerebro-vault repository. Copyright remains with the original publisher (arxiv.org).
Computer Science > Artificial Intelligence [Submitted on 14 Apr 2025 (v1), last revised 9 Jun 2026 (this version, v4)] Title:Position: Stop Anthropomorphizing Intermediate Tokens as Reasoning/Thinking Traces! View PDF HTML (experimental)Abstract:Intermediate token generation (ITG), where a model produces output before the solution, has become a standard method to improve the performance of language models on reasoning tasks. These intermediate tokens have been called \say{reasoning traces} or even \say{thinking traces} -- implicitly anthropomorphizing the traces, and implying that these
Community take
Intermediate tokens are computational scratch space, not reasoning traces—anthropomorphizing them misrepresents model internals.
Backlinks
Appeared in 1 briefing