quant_summary.json

{
  "text_encoder": {
    "mode": "copied_unchanged"
  },
  "transformer": {
    "num_shards": 7,
    "num_output_entries": 534,
    "num_fp8_quantized": 203,
    "num_weight_scales_added": 203,
    "num_non_fp8_copied": 128,
    "total_size": 32231164544
  }
}