Unverified50% confidenceFactExact time
In streaming ASR, already-emitted tokens are treated as conditioning context by subsequent decoding steps, and errors may be solidified and propagated by the language model.
1
Sources
50%
Confidence
Long-term
Relevance
10/20/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedLanguage models process tokens rather than numbers, so they can accurately copy out amounts yet are prone to errors in operations such as summation, averaging and percentages, and the result for the same question may differ from one run to the next.70% 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68% similarUnverified语言模型生成 token 本质上是内存带宽受限任务而非纯算力受限65% similarUnverifiedChanges to the tokenizer's vocabulary size, special token definitions, or context handling methods will cause some preprocessing and postprocessing logic to need updating.65% similarUnverifiedContinuous diffusion language models map discrete tokens to a continuous embedding space, perform the noising and denoising process in that space, and finally decode back to discrete tokens.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1074673API
curl https://kongchang.com/api/v1/knowledge/claims/1074673MCP
get_claim(id=1074673)