testdata/reference.json

{
  "schema_version": 1,
  "checkpoint": "tencent/Youtu-VL-4B-Instruct@8d30a0e49662a1d628a472b12df264dbcd768753",
  "image": "coco_cats.jpg",
  "image_sha256": "dea9e7ef97386345f7cff32f9055da4982da5471c48d575146c796ab4563b04e",
  "prompt": "Describe this image in one sentence.",
  "pytorch": {
    "prompt_tokens": 136,
    "image_tokens": 108,
    "generated_ids": [10392, 5240, 1896],
    "text": "Two tabby"
  },
  "ncnn_linux": {
    "generated_ids": [10392, 5240, 1896],
    "text": "Two tabby",
    "vision_max_abs": 0.0132732,
    "vision_mean_abs": 0.0000959808,
    "prefill_mode": "tokenwise_decode_graph"
  },
  "ncnn_windows": {
    "generated_ids": [10392, 5240, 1896],
    "text": "Two tabby",
    "vision_max_abs": 0.0132427,
    "vision_mean_abs": 0.0000960921,
    "prefill_mode": "tokenwise_decode_graph",
    "exit_code": 0
  }
}