reports/export-report.json

{
  "model_revision": {
    "repo_id": "genshiai-daichi/baberu-ocr",
    "revision": "d9cc13153e9a1cd8fdfa3b7b1cc329da2020aeae",
    "allow_patterns": [
      "README.md",
      "config.json",
      "configuration_baberu.py",
      "generation_config.json",
      "model.safetensors",
      "modeling_baberu.py",
      "inference.py",
      "onnx_infer.py",
      "quantization-results.md",
      "onnx/vision_fp16.onnx",
      "onnx/vision_int4.onnx",
      "onnx/decoder_prefill_int8.onnx",
      "onnx/decoder_step_int8.onnx",
      "tokenization_baberu.py",
      "tokenizer/*"
    ]
  },
  "opset": 17,
  "graphs": {
    "prefill": {
      "bytes": 133580579,
      "sha256": "3c1bc2be3e78ce90225c68b1db48d24684d7c67a3b6a26306a7ea7ff7d9452f9",
      "opsets": {
        "ai.onnx": 17
      },
      "operators": {
        "Add": 55,
        "Cast": 50,
        "Concat": 13,
        "Constant": 290,
        "ConstantOfShape": 14,
        "Div": 26,
        "Equal": 14,
        "Expand": 14,
        "Gather": 2,
        "LessOrEqual": 1,
        "MatMul": 56,
        "Mul": 138,
        "Neg": 12,
        "ReduceMean": 25,
        "Reshape": 36,
        "Sigmoid": 6,
        "Slice": 31,
        "Softmax": 6,
        "Sqrt": 25,
        "Tanh": 1,
        "Transpose": 30,
        "Unsqueeze": 16,
        "Where": 15
      },
      "inputs": {
        "vision_embeds": [
          1,
          256,
          512
        ]
      },
      "outputs": {
        "logits": [
          1,
          1,
          14630
        ],
        "present_k0": [
          1,
          2,
          257,
          64
        ],
        "present_k1": [
          1,
          2,
          257,
          64
        ],
        "present_k2": [
          1,
          2,
          257,
          64
        ],
        "present_k3": [
          1,
          2,
          257,
          64
        ],
        "present_k4": [
          1,
          2,
          257,
          64
        ],
        "present_k5": [
          1,
          2,
          257,
          64
        ],
        "present_v0": [
          1,
          2,
          257,
          64
        ],
        "present_v1": [
          1,
          2,
          257,
          64
        ],
        "present_v2": [
          1,
          2,
          257,
          64
        ],
        "present_v3": [
          1,
          2,
          257,
          64
        ],
        "present_v4": [
          1,
          2,
          257,
          64
        ],
        "present_v5": [
          1,
          2,
          257,
          64
        ]
      }
    },
    "step": {
      "bytes": 133610906,
      "sha256": "88025262335739ca5b561d55a1cbd83c6a5beda7919bc03ad3b309ad7b02714f",
      "opsets": {
        "ai.onnx": 17
      },
      "operators": {
        "Add": 57,
        "Cast": 71,
        "Concat": 69,
        "Constant": 518,
        "ConstantOfShape": 13,
        "Div": 36,
        "Equal": 13,
        "Expand": 13,
        "Gather": 71,
        "LessOrEqual": 1,
        "MatMul": 56,
        "Mul": 149,
        "Neg": 12,
        "Range": 1,
        "ReduceMean": 25,
        "Reshape": 49,
        "Shape": 82,
        "Sigmoid": 6,
        "Slice": 30,
        "Softmax": 6,
        "Sqrt": 25,
        "Sub": 1,
        "Tanh": 1,
        "Transpose": 30,
        "Unsqueeze": 180,
        "Where": 14
      },
      "inputs": {
        "token_one_hot": [
          1,
          1,
          14630
        ],
        "position_ids": [
          1,
          1
        ],
        "past_k0": [
          1,
          2,
          "past_len",
          64
        ],
        "past_k1": [
          1,
          2,
          "past_len",
          64
        ],
        "past_k2": [
          1,
          2,
          "past_len",
          64
        ],
        "past_k3": [
          1,
          2,
          "past_len",
          64
        ],
        "past_k4": [
          1,
          2,
          "past_len",
          64
        ],
        "past_k5": [
          1,
          2,
          "past_len",
          64
        ],
        "past_v0": [
          1,
          2,
          "past_len",
          64
        ],
        "past_v1": [
          1,
          2,
          "past_len",
          64
        ],
        "past_v2": [
          1,
          2,
          "past_len",
          64
        ],
        "past_v3": [
          1,
          2,
          "past_len",
          64
        ],
        "past_v4": [
          1,
          2,
          "past_len",
          64
        ],
        "past_v5": [
          1,
          2,
          "past_len",
          64
        ]
      },
      "outputs": {
        "logits": [
          "Mullogits_dim_0",
          "Mullogits_dim_1",
          14630
        ],
        "present_k0": [
          1,
          2,
          "total_len",
          64
        ],
        "present_k1": [
          1,
          2,
          "total_len",
          64
        ],
        "present_k2": [
          1,
          2,
          "total_len",
          64
        ],
        "present_k3": [
          1,
          2,
          "total_len",
          64
        ],
        "present_k4": [
          1,
          2,
          "total_len",
          64
        ],
        "present_k5": [
          1,
          2,
          "total_len",
          64
        ],
        "present_v0": [
          1,
          2,
          "total_len",
          64
        ],
        "present_v1": [
          1,
          2,
          "total_len",
          64
        ],
        "present_v2": [
          1,
          2,
          "total_len",
          64
        ],
        "present_v3": [
          1,
          2,
          "total_len",
          64
        ],
        "present_v4": [
          1,
          2,
          "total_len",
          64
        ],
        "present_v5": [
          1,
          2,
          "total_len",
          64
        ]
      }
    }
  },
  "parity": {
    "prefill": {
      "max_abs_by_output": [
        9.5367431640625e-06,
        2.384185791015625e-07,
        4.76837158203125e-07,
        1.430511474609375e-06,
        1.3709068298339844e-06,
        1.0579824447631836e-06,
        1.5497207641601562e-06,
        1.341104507446289e-07,
        1.6391277313232422e-07,
        2.2351741790771484e-07,
        1.825392246246338e-07,
        2.0116567611694336e-07,
        2.980232238769531e-07
      ],
      "max_abs": 9.5367431640625e-06,
      "top_token": 6841
    },
    "step": {
      "257": {
        "max_abs_by_output": [
          3.4809112548828125e-05,
          1.3113021850585938e-06,
          4.76837158203125e-07,
          1.430511474609375e-06,
          1.430511474609375e-06,
          1.9073486328125e-06,
          2.1457672119140625e-06,
          2.980232238769531e-07,
          1.341104507446289e-07,
          2.039596438407898e-07,
          3.5762786865234375e-07,
          1.9371509552001953e-07,
          2.551823854446411e-07
        ],
        "max_abs": 3.4809112548828125e-05,
        "top_token": 2
      },
      "288": {
        "max_abs_by_output": [
          1.4185905456542969e-05,
          9.5367431640625e-07,
          5.364418029785156e-07,
          1.1920928955078125e-06,
          1.430511474609375e-06,
          1.3113021850585938e-06,
          1.3709068298339844e-06,
          2.980232238769531e-07,
          1.1920928955078125e-07,
          1.9371509552001953e-07,
          2.682209014892578e-07,
          1.862645149230957e-07,
          3.5762786865234375e-07
        ],
        "max_abs": 1.4185905456542969e-05,
        "top_token": 5743
      }
    }
  }
}