Unverified50% confidenceFactExact time
\sysname trains a "counterfactual gain scorer" that fuses acoustic features and linguistic features, outputs token-level scores, and aggregates them into segment-level scores.
1
Sources
50%
Confidence
Long-term
Relevance
10/20/2026
First Seen
Sources
Related Entities
Related Claims
Unverified\sysname uses a pair-HMM to probabilistically align the noisy decoding results with the clean-audio decoding results, generating token-level soft supervision labels.70% similarUnverified非自回归机器翻译通过知识蒸馏解耦策略使用自回归教师模型生成的伪标签训练，将BLEU分数提升幅度超过10个点，同时推理保持并行解码64% similarUnverifiedThe authors used a byte-level neural network sandhi splitter and lemmatizer for preprocessing, then trained word vectors separately for each period.62% similarUnverifiedImproving resume matching should start with structured field extraction, transforming the matching from fuzzy semantic comparison to field-level rule-plus-semantic hybrid scoring.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1074675API
curl https://kongchang.com/api/v1/knowledge/claims/1074675MCP
get_claim(id=1074675)