# PP-OCRv6 tiny analysis inputs Source: `lw.PPOCR.Inference/models/ppocrv6-tiny` in the local PPOCR workspace. Copied files: - `det.onnx` — SHA-256 `193bab7a04fca699a6c82e6abb5b81bdb28177f0abd4062552b04908dafb19f8` - `cls.onnx` — SHA-256 `dd8b2b61983d76ab230a58da9e0e0e84956b71c3877f2ce6e438fe22d74d2cf2` - `rec.onnx` — SHA-256 `9ef676d6ed3c88256a2d92c640c44f25b0c40947e111b14b8be8f594091563e6` - `ppocr_keys.txt` — SHA-256 `46e1b34ef45684cb46d75ac76d355341fe7f0a2c38d6ee02e63ae6b3878019fc` - `sample.jpg` — SHA-256 `30c417c9f758a3b62718729f5a944f7d2e10cdd2bde0e8ce6785523ddb68ffe9` - `model.json` — model-package manifest, schema version 1 The hashes are checked by the model-analysis tests so silent asset changes fail the first development gate.