reports/optimization-embedding-w8-conversion.json

{
  "source": "build/gliner2_small_classification_fp16_L128_K8.mlpackage",
  "output": "build/gliner2_small_classification_embedding_w8_L128_K8.mlpackage",
  "method": "embedding",
  "granularity": "per_channel",
  "selected_weights": [
    "encoder_embeddings_word_embeddings_weight_to_fp16"
  ],
  "source_bytes": 151542752,
  "output_bytes": 102771476,
  "compression_seconds": 2.445869125018362
}