tripy-v0.0.9

What's Changed

• Reduce nanogpt quantization calib size to speed up test by @yizhuoz004 in #461
• Move internal change: [compiler] Add compilation task registry by @christopherbate in #465
• Disables container caching to save space by @pranavm-nvidia in #466
• Fixes a flaky test by increasing tolerance by @pranavm-nvidia in #462
• Tripy eager cache by @farazkh80 in #418
• Move internal changes by @shelkesagar29 in #468
• Support small and tiny config for SAM2 by @yizhuoz004 in #469
• Fix sam2 sample artifact removal by @yizhuoz004 in #470
• Refactors some guides by @pranavm-nvidia in #471
• Uses OpenCV package that includes binary dependencies by @pranavm-nvidia in #472
• Sandboxes notebook and example tests by @pranavm-nvidia in #473
• Move internal changes by @christopherbate in #475
• Refactors several more docs by @pranavm-nvidia in #474
• Minor docstring fixes by @pranavm-nvidia in #476
• More documentation refactoring by @pranavm-nvidia in #477
• SAM2: Save engines to different paths according to model type by @yizhuoz004 in #478
• Updates Tripy to work with Python 3.9, moves files into utils by @pranavm-nvidia in #479
• Fix engine cache path for sam2 testing by @yizhuoz004 in #481
• Refactors quantization guide by @pranavm-nvidia in #480
• Updates Tripy version to 0.0.9 by @pranavm-nvidia in #482

Full Changelog: tripy-v0.0.8...tripy-v0.0.9

tripy-v0.0.8

What's Changed

• Run L1 tests serially to prevent OOMs by @pranavm-nvidia in #454
• Updates install instructions to point to PyPI by @pranavm-nvidia in #453
• Enable contributions from external sources by @pranavm-nvidia in #456
• Move internal changes by @christopherbate in #455
• Disables pushing benchmark results for forks by @pranavm-nvidia in #458

Full Changelog: tripy-v0.0.7...tripy-v0.0.8

tripy-v0.0.7

What's Changed

• Replace tp.Parameter with tp.Tensor in resnet notebook by @yizhuoz004 in #433
• Combine the functionality of the convert_to_tensors decorator and the dtypes constraint. by @slyubomirsky in #420
• Adds explicit notebook tests, updates CI by @pranavm-nvidia in #436
• Remove plugin WAR due to mlir-tensorrt issue #915 by @parthchadha in #438
• Always construct memref value in storage op by @yizhuoz004 in #439
• Move internal changes by @shelkesagar29 in #437
• Enable sam2 video pipeline by @parthchadha in #444
• Manage peak memory in SAM2 by @parthchadha in #449
• Renames package to nvtripy to avoid name collisions on PyPI by @pranavm-nvidia in #447
• Updates version to 0.0.7 by @pranavm-nvidia in #450
• Updates release pipeline to only run L0 tests by @pranavm-nvidia in #452

Full Changelog: tripy-v0.0.6...tripy-v0.0.7

tripy-v0.0.6

What's Changed

• [mlir-tensorrt] NFC: update GitHub actions caching config for MLIR-TRT jobs by @christopherbate in #430
• Upgrade tripy version to 0.0.6 by @parthchadha in #424

Full Changelog: mlir-tensorrt-v0.1.38...tripy-v0.0.6

MLIR-TensorRT v0.1.38

mlir-tensorrt-v0.1.38

Updates nanoGPT to not store certain tensors as members

tripy-v0.0.5

What's Changed

• Add compile fixture to test integration ops with compile mode by @parthchadha in #387
• Update mlir-tensorrt dependency version in Tripy by @parthchadha in #399

Full Changelog: tripy-v0.0.4...tripy-v0.0.5

tripy-v0.0.4

What's Changed

• Update tripy version to 0.0.4 by @parthchadha in #397

Full Changelog: mlir-tensorrt-v0.1.37...tripy-v0.0.4

mlir-tensorrt-v0.1.37: [mlir-tensorrt] Add TensorRT 8.6 support (#391)

This PR makes the following changes,
    - Make TensorRT 10.5 as default version
    - Add TensorRT 8.6 download support
    - Add TensorRT 8.6 to CI
    - TensorRT 9 checks from CI are removed to deal with
    device space error.
    - Fix tests to support above changes

tripy-v0.0.3

What's Changed

• Adds a new workflow to update package index by @pranavm-nvidia in #354
• Improves various guides, hides incomplete Executable APIs by @pranavm-nvidia in #355
• Updates MLIR-TRT to 0.1.36 by @pranavm-nvidia in #356

Full Changelog: mlir-tensorrt-v0.1.36...tripy-v0.0.3

mlir-tensorrt-v0.1.36: Update version, fix minor issues (#351)

This PR makes the following changes,
- Update project version to `0.1.36`
- Torch is python dependency. Version is updated to `2.2.0+cpu` to avoid
internal CI failure
- Fix message printing issue in CI