Unverified50% confidenceBenchmarkExact time
Benchmark datasets for sign language translation (such as PHOENIX-2014T, CSL-Daily, and How2Sign) typically contain only thousands to tens of thousands of samples, far smaller than the millions of sentence pairs in parallel corpora in the machine translation field.
1
Sources
50%
Confidence
Long-term
Relevance
8/4/2026
First Seen
Sources
Hand Wave：Meta智能眼镜实时手语翻译，让听障沟通无障碍
Related Claims
UnverifiedThe context window of large language models is limited, and the length of text that can be processed in a single inference is typically tens of thousands to hundreds of thousands of tokens.72% similarUnverifiedThe vocabulary of a standard autoregressive language model typically contains 32,000 to 128,000 tokens.70% similarUnverified孟加拉语-英语的高质量平行数据仅有数百万级别，而英法语对拥有数千万句对齐语料69% similarUnverified最新的大语言模型支持12.8万到100万token的上下文，但处理数十万字长篇小说时仍会出现信息遗忘和前后矛盾69% similarUnverifiedEnd-to-end speech translation requires large amounts of aligned data pairing source-language speech with target-language text, which is extremely scarce for low-resource language pairs.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/684588API
curl https://kongchang.com/api/v1/knowledge/claims/684588MCP
get_claim(id=684588)