plapre-pico-coreml / scripts /attention.py

Commit History

  1. Performance improvements on iOS

    Daniel Rothmann
    ffa94c8
  2. Drop prefill model, add tokenizer config

    Daniel Rothmann
    95c6137
  3. Fix audio decoder, fix broken KV cache

    Daniel Rothmann
    d1bfb8c

  1. Fix a NaN attention issue in converted model

    Daniel Rothmann
    a2c97d7
  2. Add CoreML prefill and decode models

    Daniel Rothmann
    cb20bed