Learning/RSI through ngrams?

Hey gang, im wondering if you in theory could use ngrams as seen with Qwen 3.8 Flash or DS4.1 in order to dynamically train the model? Normally the ngram embeddings behave similar to a lookup table of sorts. So instead of every token having to be represented only inside the main model weights, the

中文全文翻译正在补齐，可先查看下方英文原文。

Hey gang, im wondering if you in theory could use ngrams as seen with Qwen 3.8 Flash or DS4.1 in order to dynamically train the model? Normally the ngram embeddings behave similar to a lookup table of sorts. So instead of every token having to be represented only inside the main model weights, the