v1.16.3

Changes

• e5f48ca Fixes a segfault that can occur when explicitly registering the CPU execution provider.

v1.16.2

Changes

• 3b23cea Fix usage of TensorRT execution provider, part 2

v1.16.1

Changes

• 99c5d43 Fix build on ARM targets.

v1.16.0

Changes

• d28e8ba Update to ONNX Runtime v1.16.0

v1.15.5

💖 If you find ort useful, you can now sponsor on Open Collective :)

Changes

• 092907a Add support for WASM targets

v1.15.4

Changes

• 93729f1 Fix usage of TensorRT execution provider (#87)
• 1236b52 Add TVM execution provider
• 19f6a40 Add QNN execution provider (#77)
• b3100d3 Add CANN execution provider

v1.15.3

Changes

• a4f4dae Fixes static linking on non-Windows targets.
• a4f4dae Fixes static linking with the binaries provided by supertone-inc/onnxruntime-build (#68)
• 2129797 Fix a warning message claiming ort 1.15 was actually 1.14.

v1.15.1

Changes

• 4824762 Fix compilation on Apple M1 and other non-Windows ARM64 targets with features = [ "load-dynamic" ]. (#63)
• dd6580f Fix compilation on Windows x86. (#64)

v1.15.0

Apologies for the delay.

Changes

• Upgrade to ONNX Runtime v1.15.1.
• Added support for IoBinding.
• Unified input & output tensors under one struct, Value.
• Use "unified" bindings with better platform support and less jank.
• Remove fetch-models from default features: #59
• Added support for boolean inputs: #60
• Fixed a segfault when dropping ORT tensors after session: #50
• Rework execution providers to be much easier to use.
• Added support for OpenVINO execution provider.
• Added support for loading models with in-memory external initializers (can be used to load LoRAs, for instance)

v1.14.6

Changes

• b6bbcd5 Adds support for global thread pools. (#43)
• e23773f Add an option to enable the CoreML execution provider only for devices with an Apple Neural Engine.
• 481d025 Makes ORT_DYLIB_PATH actually relative to the executable like the README says.
• d4b9972 Logs a warning when attempting to register execution providers that ort was not compiled to support.
• 6f0f746 Fix overly verbose debug messages when dropping tensors.
• 126c742 Improve logging messages with better descriptions.
• e79ceea Fix broken links and missing documentation.