5 projects
codecscope
Measure how efficiently neural audio codecs tokenize audio: token rate, bitrate, codebook utilization, and reconstruction quality across DAC, EnCodec, SNAC, and Mimi
sinhaladate
A Python library for parsing and formatting Sinhala dates and times
cetana
Auditable BDI (belief-desire-intention) agents with LLM deliberation — explicit beliefs, committed intentions, and a trace that answers 'why'
tokscope
Measure how efficiently LLM tokenizers handle your language: fertility, UNK rate, and token premium across HuggingFace, tiktoken, and SentencePiece tokenizers
slnic
Parse, validate, convert, and generate Sri Lankan NIC (National Identity Card) numbers
