Unverified50% confidenceFactExact time
TinyLlama-1.1B's tokenizer produces an average of 1.34 subword tokens per word when processing Latin script, and an average of 22.60 tokens per word when processing Kannada.
1
Sources
50%
Confidence
Long-term
Relevance
10/20/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedFertility is calculated as the total number of tokens produced by the tokenizer after processing the target-language corpus divided by the total number of original words.72% similarUnverifiedIn TinyLlama's vocabulary, the codepoint coverage is 0% for Odia, 1% for Telugu, and 2% for Kannada.72% similarUnverifiedRico Sennrich et al. quantified in their 2016 paper "Neural Machine Translation of Rare Words with Subword Units": on a 100-million-word German corpus, keeping whole words requires a vocabulary of 1.75 million entries, and the test set still leaves 1,079 unknown words; splitting into characters requires about 3,000 symbols, but the corpus balloons to 550 million positions; the subword method BPE uses about 63,000 symbols, 112 million positions, and zero unknown words.71% similarUnverifiedFor code, each line consumes approximately 10–15 tokens on average, which is higher than the rate of 1.3 tokens per word for natural language text.70% similarUnverified对于英文代码1个token大约对应4个字符，100万token的上下文窗口大约等同于75万个英文单词70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1074683API
curl https://kongchang.com/api/v1/knowledge/claims/1074683MCP
get_claim(id=1074683)