model_info:
  name: anemll-google-gemma-3-1b-it-ctx512
  version: 0.3.5
  description: |
    Demonstarates running google-gemma-3-1b-it on Apple Neural Engine
    Context length: 512
    Batch size: 64
    Chunks: 2
  license: MIT
  author: Anemll
  framework: Core ML
  language: Python
  architecture: gemma3_text
  parameters:
    context_length: 512
    batch_size: 64
    lut_embeddings: none
    lut_ffn: none
    lut_lmhead: none
    num_chunks: 2
    model_prefix: gemma3
    embeddings: gemma3_embeddings.mlmodelc
    lm_head: gemma3_lm_head.mlmodelc
    ffn: gemma3_FFN_PF_chunk_01of02.mlmodelc
    split_lm_head: 16
    argmax_in_model: true
    sliding_window: 512
    prefill_dynamic_slice: true
    single_cache: true