Unverified50% confidenceFactExact time
Transformer inference is divided into two stages, prefill and decode: prefill processes the entire input prompt in parallel at once and is mainly compute-bound; decode generates output tokens one by one and is mainly memory-bandwidth-bound.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedP/D Disaggregation (Prefill/Decode Disaggregation) deploys the Prefill and Decode stages on different nodes to independently scale according to their respective computational characteristics, while simultaneously optimizing throughput and time to first token (TTFT).77% similarVerifiedThe Transformer's encoder is responsible for mapping the input sequence to a contextual representation and is suitable for classification and understanding tasks, while the decoder autoregressively generates the output sequence and is suitable for text generation tasks77% similarUnverifiedTransformer uses Positional Encoding to retain word order information while performing parallel computation.76% similarUnverified输入Token可以在prefill阶段并行处理，而输出Token必须在decode阶段逐个串行生成76% similarVerifiedThe original Transformer consists of an Encoder and a Decoder; BERT-type models use only the Encoder and excel at understanding and classification, while GPT-type models use only the Decoder and focus on text generation.76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1046820API
curl https://kongchang.com/api/v1/knowledge/claims/1046820MCP
get_claim(id=1046820)