$ ./ymmv --about

  - Integer-native training for transformer language models
  - Weights and optimizer state as 8-bit integers from end to end

$ ./ymmv --why

  - Serve-time iso-quality artifact deploy-ready with no PTQ
  - Drastically reduced memory footprint allows training models
    that otherwise cannot load at all on the given hardware

$ ./ymmv --status

  - Working proof of concept: GPT-2 and Llama 3.2 families
  - Patent pending

$ ./ymmv --contact

  - joseph@algorithm-intelligence.ai
  - (c) 2026 Algorithm Intelligence LLC

$