Fix q8 weights (use uint8 for q8; int8 produces poor results) (#18)

- Upload fixed q8 ONNX models (reduce_range=True, per_channel=True) (06633a3e0bcdf1c31bd8ce0a27cb764aff26d6a9)
- Upload folder using huggingface_hub (0919b6ca05a1e2b9e0305936f0d125a688274007)
- Fix q8 weights (use uint8 for q8; int8 produces poor results) (4f13109a9d4babf8776d1a4f860c3634f28a4be3)

onnx/model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f8eeead8e191939562a98af969f9d63dd404dc72a9a65b2c19cabb857de7c8d9
-size 1714133062
+oid sha256:6e038fd6fb27b41fbb62e6a7df9b60b57215db3958d14382221beaab78fbc1d4
+size 1714133130