Unverified50% confidenceFactExact time
自回归生成源自2017年Transformer架构的提出，模型每次只生成一个词元，并将已生成的所有词元作为下一步的上下文输入
1
Sources
50%
Confidence
Long-term
Relevance
7/17/2026
First Seen
Sources
Related Claims
UnverifiedTransformer-based LLMs perform an independent forward-pass computation for each inference, with model weights fixed during the inference phase; what is referred to as context memory is essentially the re-injection of historical conversation text into the prompt.74% similarUnverifiedCurrent mainstream large language models adopt the autoregressive generation mechanism of the Transformer architecture, generating token by token from left to right, and cannot go back to modify already-output reasoning steps after generation.74% similarVerifiedLLMs generate text in an autoregressive manner based on the Transformer architecture, predicting only one token per forward pass and then appending that token to the input sequence before predicting the next token.74% similarUnverifiedTransformer架构在每次推理时必须处理完整的输入序列，代理执行到第N步时输入上下文必须包含前N-1步的完整执行轨迹73% similarUnverifiedEach additional token generated by an autoregressive model provides one more opportunity for a Transformer forward pass computation, trading sequence length for computational depth.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/536671API
curl https://kongchang.com/api/v1/knowledge/claims/536671MCP
get_claim(id=536671)