BitsExplorer

BitsExplorer finds what people are building with AI, every day. See today's picks →

Preview of Thai Tokenizer Retrofit

Thai Tokenizer Retrofit

Cut Thai token cost at fixed vocabulary size: prune unused entries and swap in Thai subwords. Mistral-7B: Thai tokens −61–67%, English unchanged.

Description excerpted from the original listing, which is linked below.

Open it Source code

The week in AI-built software

Every Monday: how many projects were built with AI that week, which tools built them, where they came from, and the week's picks.

Read the latest week