100 Pull requests merged by 41 people

• Event Tracer Constraint

  #13521 merged Aug 23, 2025

• Call .detach() in static attention cache update helper

  #13618 merged Aug 23, 2025

• fix MM nullptr from zero bias

  #13523 merged Aug 23, 2025

• migrate all test_aten_ops to facto

  #13483 merged Aug 23, 2025

• Split on depthwise for strongly typed convs

  #13562 merged Aug 23, 2025

• Set an empty EValue input for models that expect None arg.

  #13621 merged Aug 23, 2025

• Add get_output API.

  #13610 merged Aug 22, 2025

• Create stale.yml workflow to label stale PRs

  #13565 merged Aug 22, 2025

• Add set_outputs() API.

  #13609 merged Aug 22, 2025

• Make TensorPtr constructor check the data dize matches the shape.

  #13591 merged Aug 22, 2025

• Access Method directly from TrainingModule.

  #13602 merged Aug 22, 2025

• Make IOManager use Module instead of Method.

  #13542 merged Aug 22, 2025

• Add missing backslashes in example run section

  #13603 merged Aug 22, 2025

• Update tokenizer to include tekken implementation

  #13601 merged Aug 22, 2025

• Summary: Add ExecutorchRuntimeException: Throw relevant exceptions from JNI layer in the event of errors

  #13526 merged Aug 22, 2025

• Unbreak build-benchmark-app (apple) after pin bump in #13334

  #13582 merged Aug 21, 2025

• Input position accessor for static attention IO manager

  #13561 merged Aug 21, 2025

• Fix memory planning for greed with heuristic algo.

  #13541 merged Aug 21, 2025

• Update test_remove_unused_parameters_pass.py

  #13563 merged Aug 21, 2025

• Non-fatal error when ET_SWITCH encounters unsupported dtype

  #13359 merged Aug 21, 2025

• Whisper audio processor

  #13538 merged Aug 21, 2025

• Add support for strongly typed quantized_op_add

  #13531 merged Aug 21, 2025

• [Backend Tester] Add markdown summary in CI

  #13535 merged Aug 20, 2025

• [Backend Tester] Add additional quantized test flows for XNNPACK and Vulkan

  #13534 merged Aug 20, 2025

• [Backend Tester] Add test flows for QNN

  #13469 merged Aug 20, 2025

• Fix method meta error in wasm build

  #13496 merged Aug 20, 2025

• Fix sigmoid operator to support boolean tensor inputs

  #13515 merged Aug 20, 2025

• Split on dilation for strongly typed convs

  #13532 merged Aug 20, 2025

• Link QNN backend to pybinding lib when built

  #13467 merged Aug 20, 2025

• Bump the PyTorch pin to 20250811

  #13334 merged Aug 20, 2025

• Remove unused sleef.h which breaks cross-compilation on windows

  #13079 merged Aug 20, 2025

• Fix buck cquery //extension/llm/runner: in OSS

  #13527 merged Aug 20, 2025

• Temporarily disable test-models-arm-zephyr

  #13548 merged Aug 20, 2025

• Arm backend: Added VGF minimal example

  #13545 merged Aug 20, 2025

• Arm backend: Update examples/arm/README.md

  #13546 merged Aug 20, 2025

• Arm backend: Add limited support for fish shell in setup_path.sh

  #13323 merged Aug 20, 2025

• Adding smollm2 to examples/models/__init__.py

  #13514 merged Aug 20, 2025

• Arm backend: Fix for combo neg(x)+1 + tests

  #13517 merged Aug 20, 2025

• Arm backend: Remove get_output_nodes from runner_utils.

  #13417 merged Aug 20, 2025

• QNN Llama Runner implement IRunner

  #13171 merged Aug 20, 2025

• Re-enable model tests with recipes for xnnpack backend

  #13519 merged Aug 20, 2025

• Fix Olmo trunk test and skip XNNPack trunk tests on mac

  #13528 merged Aug 20, 2025

• Update project.pbxproj

  #13537 merged Aug 20, 2025

• Refactor pybinding unit test

  #13479 merged Aug 20, 2025

• Remove pinning to ET commit for Zephyr CI job

  #13388 merged Aug 20, 2025

• Qualcomm AI Engine Direct - GA Static Smollm2

  #13406 merged Aug 19, 2025

• [Backend Tester] Add QNN test flow to nightly CI

  #13468 merged Aug 19, 2025

• Qualcomm AI Engine Direct - Remove input_list dependencies

  #13411 merged Aug 19, 2025

• Fix buck query //extension/flat_tensor: in OSS

  #13484 merged Aug 19, 2025

• Remove other extensions' source files from EXTENSION_TRAINING_SRCS and TRAIN_XOR_SRCS

  #13433 merged Aug 19, 2025

• Improve softmax perf when transpose is not needed

  #13081 merged Aug 19, 2025

• [Backend Tester] Run Core ML tests in nightly CI

  #13446 merged Aug 19, 2025

• [Backend Tester] Run Vulkan tests in nightly CI

  #13445 merged Aug 19, 2025

• [Backend Tester] Mark adaptive avgpool2d as an unsupported portable op

  #13398 merged Aug 19, 2025

• removed lines of cron schedule runs until the github token issue is fixed

  #13486 merged Aug 19, 2025

• Remove unused PROGRAM_SCHEMA_SRCS from build_variables.bzl

  #13432 merged Aug 19, 2025

• Add Model Profiling Automation Script

  #13493 merged Aug 19, 2025

• [Backend Tester] Add nightly CI job for XNNPACK

  #13390 merged Aug 19, 2025

• Delete extract_sources.py and cmake_deps.toml

  #13395 merged Aug 19, 2025

• Wrap LLM runner tests in anonymous namespace (unbreak unittest-release)

  #13524 merged Aug 19, 2025

• Qwen and Phi-4-Mini targets

  #13449 merged Aug 19, 2025

• Clean up apparently-unnecessary mentions of BUCK2 in scripts

  #13394 merged Aug 19, 2025

• Remove NTSTATUS cast

  #9510 merged Aug 19, 2025

• Force -O3 for executorch op_div.cpp in clang 19

  #13480 merged Aug 19, 2025

• Add a generic multimodal runner

  #13166 merged Aug 19, 2025

• Summary: Add MCU model script to validate and run the models

  #13439 merged Aug 19, 2025

• Stop looking for buck2 in the top-level ExecuTorch build

  #13393 merged Aug 19, 2025

• Only support int8 and quant dtypes for quant operators

  #11685 merged Aug 19, 2025

• Stop validating that build_variables.bzl matches buck-generated executorch_srcs.cmake

  #13392 merged Aug 19, 2025

• Reset Temp Allocator after each use

  #13384 merged Aug 19, 2025

• Arm backend: Add support for QAT+per-channel combo

  #13511 merged Aug 19, 2025

• [ET-VK][ez] Fix erroneous cherry-pick bot merge

  #13512 merged Aug 19, 2025

• [ET-VK] Allocate memory for weight and activation tensors lazily

  #13501 merged Aug 19, 2025

• [ET-VK][AOT] Serialize constant tensors via NamedDataMap

  #13499 merged Aug 19, 2025

• [ET-VK] Runtime support for NamedDataMap

  #13498 merged Aug 19, 2025

• [ET-VK][ez] Move execute node threshold calculation from prepare_pipelines() to prepare()

  #13497 merged Aug 19, 2025

• [ET-VK] Allocate memory for weight and activation tensors lazily

  #13474 merged Aug 19, 2025

• [ET-VK][AOT] Serialize constant tensors via NamedDataMap

  #13473 merged Aug 19, 2025

• [ET-VK] Runtime support for NamedDataMap

  #13472 merged Aug 19, 2025

• [ET-VK][ez] Move execute node threshold calculation from prepare_pipelines() to prepare()

  #13478 merged Aug 19, 2025

• Arm backend: Update tosa dialect buck file

  #13487 merged Aug 18, 2025

• Fix test-binary-size-linux -Wsign-compare failure with c10::irange

  #13481 merged Aug 18, 2025

• Enable BNNS copy for FP16 to FP32

  #13477 merged Aug 18, 2025

• Switch non-top-level ExecuTorch builds (size test, examples, etc.) from executorch_srcs.cmake to build_variables.bzl

  #13391 merged Aug 18, 2025

• Add a default image prefiller implementation

  #13310 merged Aug 18, 2025

• Add torchao to swiftpm

  #13401 merged Aug 18, 2025

• NXP backend: Improve target support checks.

  #13367 merged Aug 18, 2025

• Remove outdated NCHW to NHWC pass and rename the current one to ReplaceConvWithChannelLastConvPass

  #13420 merged Aug 18, 2025

• NXP backend: Remove optimization in prune_reshape_operators.py

  #13413 merged Aug 18, 2025

• NXP backend: Remove optimization in prune_cast_operators.py

  #13377 merged Aug 18, 2025

• NXP backend: Remove optimization in fuse_quanitze_into_preceding_ops.py

  #13372 merged Aug 18, 2025

• Arm backend: Add partial vulkan runtime support for VgfPipeline

  #13471 merged Aug 18, 2025

• Arm backend: Add cumsum support

  #13457 merged Aug 18, 2025

• Arm backend: Add example linkerscripts for U55/U85

  #13404 merged Aug 18, 2025

• Arm backend: Move TOSA operators to dialect

  #13408 merged Aug 18, 2025

• Arm backend: Replace .export_for_training with .export

  #13280 merged Aug 18, 2025

• Arm backend: Add dim_order_ops:: to the auto gen_oplist generations

  #13455 merged Aug 18, 2025

• Add support for strongly typed conv_nchw and conv_nhwc

  #13462 merged Aug 16, 2025

• Add aten::_upsample_bilinear2d_aa.out

  #13458 merged Aug 16, 2025

• [ET-VK] Enable IntxWeightOnlyConfig

  #13466 merged Aug 16, 2025

51 Pull requests opened by 29 people

• add ability to compare intermedidate outputs

  #13482 opened Aug 18, 2025

• Add check_for_installed_private_headers_in_cmake_out

  #13485 opened Aug 18, 2025

• NXP backend: Refactor executorch_pipeline.py, allow different sources for calibration inputs

  #13494 opened Aug 18, 2025

• [ET-VK] Allocate memory for weight and activation tensors lazily

  #13500 opened Aug 19, 2025

• Qualcomm AI Engine Direct - Add&update README

  #13504 opened Aug 19, 2025

• NXP backend: Add support for the `aten.cat` operator.

  #13505 opened Aug 19, 2025

• NXP backend: Add implementation of Tanh operator converter

  #13510 opened Aug 19, 2025

• Update deprecated pkg_resources api in edge._ops

  #13513 opened Aug 19, 2025

• Arm backend: Annotate ADD/SUB with indepenedent observers

  #13516 opened Aug 19, 2025

• Qualcomm Whisper Runner JNI bindings

  #13525 opened Aug 19, 2025

• Fix remaining instances of `for aten in (True, False):`

  #13529 opened Aug 19, 2025

• Bump optimum ET pin

  #13530 opened Aug 19, 2025

• Add 1MB temp allocator for executor

  #13533 opened Aug 19, 2025

• StaticAttentionIOManager: support attention sink

  #13540 opened Aug 20, 2025

• Let's not expose the underlying Method.

  #13543 opened Aug 20, 2025

• Qualcomm AI Engine Direct - Scripts and accuracy improvement for Qwen3_0.6B/1.7B and Qwen 2.5_1.5B

  #13544 opened Aug 20, 2025

• NXP backend: Add support for conversion of Conv1D operator

  #13549 opened Aug 20, 2025

• Cortex_m backend: Loosen edge op check.

  #13550 opened Aug 20, 2025

• install headers from runtime/backend

  #13558 opened Aug 20, 2025

• Switch from print_exported_headers to CMake's installed headers in scripts/build_apple_frameworks.sh

  #13559 opened Aug 20, 2025

• [Core ML] Improve asset management

  #13560 opened Aug 20, 2025

• Fix data loader build on Windows, re-enable pybind job

  #13564 opened Aug 20, 2025

• Added JS bindings for tokenizers library

  #13566 opened Aug 21, 2025

• Run Python unit test CI on Windows

  #13567 opened Aug 21, 2025

• Add 16A8W support and test for add operation

  #13568 opened Aug 21, 2025

• Add 16A8W support and test for sigmoid operation

  #13569 opened Aug 21, 2025

• Add 16A8W support and test for tanh operation

  #13570 opened Aug 21, 2025

• Summary: Add MCU model verification to CI

  #13572 opened Aug 21, 2025

• Qualcomm AI Engine Direct - Improve GA Static Phi-4-mini accuracy

  #13573 opened Aug 21, 2025

• NXP backend: Remove IR optimization to remove dead branches.

  #13574 opened Aug 21, 2025

• Arm backend: Dont try to fuse const for TOSA ops

  #13575 opened Aug 21, 2025

• NXP backend: Use zero point for quantized padding.

  #13576 opened Aug 21, 2025

• Arm backend: Fix nummerical accuracy error.

  #13578 opened Aug 21, 2025

• Add pybindings for attribute tensors

  #13579 opened Aug 21, 2025

• Delete tools/cmake/{buck_util,resolve_buck}.py

  #13588 opened Aug 21, 2025

• Override unload_method in training_module to erase the tensors pointing to the released memory

  #13590 opened Aug 21, 2025

• Inline requantize kernels

  #13592 opened Aug 22, 2025

• [ET-VK][testing] Add scripts to facilitate operator testiing

  #13593 opened Aug 22, 2025

• [ET-VK][ez] Consolidate tensor metadata calculation + buffer binding code

  #13594 opened Aug 22, 2025

• [ET-VK] Introduce `BufferMetadata` GLSL struct to abstract tensor layout

  #13595 opened Aug 22, 2025

• [ET-VK][ez] Allow high dimensional tensors (for buffer storage)

  #13596 opened Aug 22, 2025

• [ET-VK] High dim tensor support for view, unsqueeze, squeeze, clone

  #13597 opened Aug 22, 2025

• [ET-VK][ez] Fix partitioner logic of finding keepdim arg of reduce ops

  #13598 opened Aug 22, 2025

• [ET-VK][ez] Support grouped convolutions

  #13599 opened Aug 22, 2025

• [ET-VK][ez] Use XNNPACK's FuseBatchNorm pass

  #13600 opened Aug 22, 2025

• Update cpuinfo pin to latest

  #13605 opened Aug 22, 2025

• Move undelegated constants

  #13606 opened Aug 22, 2025

• Enumerated shape

  #13613 opened Aug 22, 2025

• Update coremltools to 9b1

  #13614 opened Aug 22, 2025

• Rename stale to stale.yml

  #13619 opened Aug 22, 2025

• Add support for data path in iOS

  #13620 opened Aug 22, 2025

15 Issues closed by 10 people

• ARM repo gitlab is not stable

  #13557 closed Aug 21, 2025

• [v0.7.0] Release Tracker

  #11075 closed Aug 21, 2025

• ExecuTorch execution failed for sigmoid.default

  #13492 closed Aug 20, 2025

• Quantization of mobilenet_v2 on an android device shows no accleration

  #13502 closed Aug 20, 2025

• Support QAT Quantization in executorch export

  #13099 closed Aug 19, 2025

• Does method_meta exist in python runtime api?

  #7709 closed Aug 19, 2025

• Add pybindings for program.h/cpp and method.h/cpp

  #11567 closed Aug 19, 2025

• Support GraphModule and ExportedProgram including nn.module

  #13104 closed Aug 19, 2025

• Add CoreML Quant Recipes

  #13263 closed Aug 19, 2025

• Update Partitioner Dtype Logic

  #11581 closed Aug 19, 2025

• [Vulkan] Reduce peak memory usage when loading models with ET-VK

  #13475 closed Aug 19, 2025

• Quantized Add Op with CMSIS-NN integration

  #13506 closed Aug 19, 2025

• [Vulkan] Improve LLM Prefill Performance

  #12920 closed Aug 18, 2025

• Freeing memory using free() which was not allocated by malloc()

  #11395 closed Aug 18, 2025

• kernel 'aten::_upsample_bilinear2d_aa.out' not found.

  #7031 closed Aug 16, 2025

38 Issues opened by 12 people

• [Android][Kotlin] Move ExecuTorch Java API to Kotlin API

  #13617 opened Aug 22, 2025

• [Android][E2E] Add E2E test for SmolLM

  #13616 opened Aug 22, 2025

• [Android][E2E] Add E2E test for Qwen3

  #13615 opened Aug 22, 2025

• QNN Avg and Max pool with ceil_mode=True appear to have incorrect outputs

  #13612 opened Aug 22, 2025

• QNN Transposed Conv2d with Dilation appears Incorrect

  #13611 opened Aug 22, 2025

• All-dim reduction operators fail with “IndexError: tuple index out of range”

  #13608 opened Aug 22, 2025

• "How to Support a Custom Model in HTP Backend" example code is out of date

  #13607 opened Aug 22, 2025

• Unsupported node types fail to lower with “KeyError”

  #13604 opened Aug 22, 2025

• ExecuTorch execution failed for div.Tensor

  #13589 opened Aug 21, 2025

• Missing Half/Bfloat16 support in portable kernels

  #13587 opened Aug 21, 2025

• ExecuTorch kernels for native_batch_norm ops do not support mixing float16/bfloat16 input and float32 params

  #13586 opened Aug 21, 2025

• ExecuTorch execution failed for masked_select.default

  #13585 opened Aug 21, 2025

• Fix whisper export

  #13584 opened Aug 21, 2025

• Android GA: Add Exceptions in JNI layer, Establish the foundational pieces

  #13583 opened Aug 21, 2025

• ExecuTorch execution failed for arange.start_step

  #13581 opened Aug 21, 2025

• ExecuTorch execution failed for arange.default

  #13580 opened Aug 21, 2025

• [v1.0.0] Release Tracker

  #13577 opened Aug 21, 2025

• Test Windows wheel build

  #13571 opened Aug 21, 2025

• ExecuTorch execution crashed for narrow_copy.default

  #13556 opened Aug 20, 2025

• ExecuTorch execution crashed for elu.default

  #13555 opened Aug 20, 2025

• ExecuTorch execution crashed for constant_pad_nd.default

  #13554 opened Aug 20, 2025

• ExecuTorch execution crashed for _upsample_bilinear2d_aa.default

  #13553 opened Aug 20, 2025

• ExecuTorch execution crashed for unbind_copy.int

  #13552 opened Aug 20, 2025

• ExecuTorch execution crashed for _log_softmax.default

  #13551 opened Aug 20, 2025

• Reenable test-models-arm-zephyr

  #13547 opened Aug 20, 2025

• Support for gradients to user inputs in TrainingModule

  #13522 opened Aug 19, 2025

• Whisper Runtime Failure

  #13520 opened Aug 19, 2025

• NXP: Enable mypy python type checker

  #13518 opened Aug 19, 2025

• Missing grid_sample_2d portable kernel

  #13509 opened Aug 19, 2025

• Zephyr : Path to GA

  #13508 opened Aug 19, 2025

• [CortexM] Stateful Ops (Linera) with CMSIS-NN integration with unit Testcases

  #13507 opened Aug 19, 2025

• Vulkan get_shader_info -- Could not find ShaderInfo with name

  #13503 opened Aug 19, 2025

• ExecuTorch execution failed for hardtanh.default

  #13491 opened Aug 18, 2025

• Convolution portable kernel missing 5D support

  #13490 opened Aug 18, 2025

• ExecuTorch execution failed for any.dims

  #13489 opened Aug 18, 2025

• ExecuTorch execution failed for add.Tensor

  #13488 opened Aug 18, 2025

• [Vulkan] Dynamically quantized 8-bit convolutions

  #13476 opened Aug 18, 2025

• Use padding=='same' for CifarNet model

  #13470 opened Aug 18, 2025

110 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Summary: Initial CMSS-NN Add Op

  #13296 commented on Aug 21, 2025 • 22 new comments

• Summary: Add Initial support to pico2 (Arm Cortex M)

  #12518 commented on Aug 20, 2025 • 21 new comments

• [EXIR] Register _clone_dim_order op and map aten.clone

  #12971 commented on Aug 22, 2025 • 17 new comments

• Add 16A8W linear ops support and test

  #13448 commented on Aug 23, 2025 • 9 new comments

• Add logical and op

  #13342 commented on Aug 22, 2025 • 6 new comments

• Add XNNPACK backend option for workspace sharing

  #11748 commented on Aug 19, 2025 • 5 new comments

• Fix aten.amax lowering issue

  #13381 commented on Aug 22, 2025 • 4 new comments

• Qualcomm AI Engine Direct - Avoid duplicated external ION memory registration

  #13421 commented on Aug 20, 2025 • 2 new comments

• Add initial backends/cadence/vision module scaffold with optimized softmax kernel (no iDMA)

  #12480 commented on Aug 21, 2025 • 2 new comments

• Arm backend: Add decomposition and test for fmod

  #13456 commented on Aug 19, 2025 • 1 new comment

• [EXIR] Update RemoveCloneOpsTransform to be dim order aware

  #12976 commented on Aug 16, 2025 • 1 new comment

• guard_or_false in dim_order == 0 check

  #12651 commented on Aug 18, 2025 • 1 new comment

• Fix XNNPACK build on Windows

  #13259 commented on Aug 21, 2025 • 1 new comment

• NXP backend: Better handling of arg indexes in Neutron quantizer

  #13368 commented on Aug 18, 2025 • 1 new comment

• Arm backend: Bump Arm baremetal compilers to gcc 14.3

  #13317 commented on Aug 20, 2025 • 1 new comment

• Elaborate on the release notes labels for backends

  #11470 commented on Aug 20, 2025 • 0 new comments

• Enforce clang for Windows builds

  #11159 commented on Aug 18, 2025 • 0 new comments

• BugFix: Modify TPS metric calculation. Add default cpu threads for hybrid CPU system.

  #11063 commented on Aug 21, 2025 • 0 new comments

• Qualcomm AI Engine Direct - GA Qwen 2.5 0.5B

  #12333 commented on Aug 21, 2025 • 0 new comments

• Arm backend: Add dump_delegate_data function

  #12334 commented on Aug 20, 2025 • 0 new comments

• Use ::std::string_view

  #12706 commented on Aug 18, 2025 • 0 new comments

• [pytorch hash update] update the pinned pytorch hash

  #10955 commented on Aug 23, 2025 • 0 new comments

• insert_write_back_for_buffers_pass should inject copy_ nodes at the earliest possible spot.

  #7345 commented on Aug 21, 2025 • 0 new comments

• GA Models : MCU - Add initial CI pipeline for MV2/MV3 (with no delegation, but quantized) on FVP

  #13405 commented on Aug 21, 2025 • 0 new comments

• Add Voxtral / general audio multimodal support

  #12714 commented on Aug 21, 2025 • 0 new comments

• Update selective build example for top-level targets

  #12948 commented on Aug 21, 2025 • 0 new comments

• AOTI backend support

  #13188 commented on Aug 21, 2025 • 0 new comments

• Shared memory for multiple entry points with XNNPACK delegate

  #11738 commented on Aug 21, 2025 • 0 new comments

• Allow delegates to hook into memory planning

  #13329 commented on Aug 21, 2025 • 0 new comments

• Prototype unified api

  #12549 commented on Aug 21, 2025 • 0 new comments

• Migrate Pybindings to Module API

  #11297 commented on Aug 21, 2025 • 0 new comments

• Installable top-level CMake targets

  #12949 commented on Aug 21, 2025 • 0 new comments

• Set up Windows CI

  #12962 commented on Aug 21, 2025 • 0 new comments

• support etdump generation in executorch.runtime

  #13464 commented on Aug 20, 2025 • 0 new comments

• fix mismatch sub dtype

  #13447 commented on Aug 22, 2025 • 0 new comments

• [Backend Tester] Upload test report artifacts in CI job

  #13444 commented on Aug 19, 2025 • 0 new comments

• Zephyr SDK Version Bump v0.17.2->v0.17.3

  #13426 commented on Aug 20, 2025 • 0 new comments

• Add OOTB tests for macos-m1-stable runner

  #13416 commented on Aug 19, 2025 • 0 new comments

• NXP backend: Remove optimization in remove_unused_tensors_and_buffers.py

  #13415 commented on Aug 18, 2025 • 0 new comments

• NXP backend: Remove optimization in prune_quantize_operators.py

  #13409 commented on Aug 18, 2025 • 0 new comments

• NXP backend: Remove optimization in keep_one_empty_buffer.py

  #13376 commented on Aug 20, 2025 • 0 new comments

• Arm backend: Update VelaIO handling

  #13282 commented on Aug 19, 2025 • 0 new comments

• Fix portable library build on Windows

  #13260 commented on Aug 21, 2025 • 0 new comments

• Create Windows CMake preset

  #13257 commented on Aug 21, 2025 • 0 new comments

• Fix devtools CMake build failure on Windows

  #13251 commented on Aug 21, 2025 • 0 new comments

• Fix error reporting in Windows preset build job

  #13247 commented on Aug 21, 2025 • 0 new comments

• [KleidiAI] turn on SME2 by default

  #13233 commented on Aug 21, 2025 • 0 new comments

• Add AOTI Lowering Recipe

  #13198 commented on Aug 21, 2025 • 0 new comments

• Add 16A8W quantization configuration utility for ARM backend

  #13175 commented on Aug 22, 2025 • 0 new comments

• NXP backend: Bump Neutron Converter SDK version to 25.06

  #13128 commented on Aug 19, 2025 • 0 new comments

• Add AOTI backend skeleton code

  #13123 commented on Aug 22, 2025 • 0 new comments

• Correct the flatc/flatcc compilation for x86 ios simulator

  #13080 commented on Aug 22, 2025 • 0 new comments

• Qualcomm AI Engine Direct - fix typo

  #13030 commented on Aug 20, 2025 • 0 new comments

• Arm backend: Make TOSA backend NCHW-compatible

  #12994 commented on Aug 22, 2025 • 0 new comments

• Use std::optional

  #12707 commented on Aug 18, 2025 • 0 new comments

• PT2 Archive weight format

  #13307 commented on Aug 21, 2025 • 0 new comments

• [RFC] Objective-C/Swift APIs for ExecuTorch

  #8360 commented on Aug 21, 2025 • 0 new comments

• Add Vulkan Delegate Benchmarks for Llama 3.2 (1B and 3B) and MobileNet

  #8293 commented on Aug 21, 2025 • 0 new comments

• Move examples + tests to single-step CMake build

  #8274 commented on Aug 21, 2025 • 0 new comments

• Improve common error messages

  #8273 commented on Aug 21, 2025 • 0 new comments

• DevX: Track the benchmark infra health and usage

  #8247 commented on Aug 21, 2025 • 0 new comments

• Document iOS Version Requirement for Core ML Delegate

  #8264 commented on Aug 21, 2025 • 0 new comments

• Support scheduling benchmark jobs via scripts

  #8245 commented on Aug 21, 2025 • 0 new comments

• Enable workflow time profiling

  #8244 commented on Aug 21, 2025 • 0 new comments

• Enable framegraphs for benchmarking on iOS

  #8243 commented on Aug 21, 2025 • 0 new comments

• Enable framegraphs for benchmarking on Android

  #8242 commented on Aug 21, 2025 • 0 new comments

• Integrate mem profiler to benchmark infra

  #8241 commented on Aug 21, 2025 • 0 new comments

• Integrate ETdump to benchmark infra

  #8240 commented on Aug 21, 2025 • 0 new comments

• Support auto regression detection and bisect

  #8239 commented on Aug 21, 2025 • 0 new comments

• Inconsistent logo between GitHub repo and documentation site

  #8163 commented on Aug 21, 2025 • 0 new comments

• Multiple Entrypoints and Shared State implementation; Request for comments/feedback.

  #8030 commented on Aug 21, 2025 • 0 new comments

• kernel 'executorch_prim::et_view.default' not found.

  #7459 commented on Aug 21, 2025 • 0 new comments

• double free or corruption (out) on running Llava on x86_64

  #4869 commented on Aug 21, 2025 • 0 new comments

• Support for Shared Weights between multiple entry points in PTE export

  #12804 commented on Aug 21, 2025 • 0 new comments

• [Sprint 4] Make Weight Sharing Default

  #12503 commented on Aug 20, 2025 • 0 new comments

• MediaTek backend failed to generate .pte

  #13361 commented on Aug 20, 2025 • 0 new comments

• Incorrect Observer Sharing/Derivation at Conv-ReLU+ Residual with Arm Ethos Quantizer

  #12959 commented on Aug 20, 2025 • 0 new comments

• Fails to import EthosUPartitioner for ARM backend as there is numpy inconsistency with tosa-tools

  #12991 commented on Aug 20, 2025 • 0 new comments

• run mtk_build_examples.sh error

  #11999 commented on Aug 20, 2025 • 0 new comments

• Cortex-M Quantization Operators Produce Incorrect Results

  #13399 commented on Aug 19, 2025 • 0 new comments

• How to run a executorch model directly from memory instead of saving it as a disk file

  #12749 commented on Aug 18, 2025 • 0 new comments

• [Vulkan] Improve latency for non-LLM models

  #12921 commented on Aug 18, 2025 • 0 new comments

• --skip-fvp-setup flag is ignored

  #12447 commented on Aug 18, 2025 • 0 new comments

• Lora Demo

  #12960 commented on Aug 21, 2025 • 0 new comments

• [Android] Add method metadata report API in Module.java

  #10453 commented on Aug 21, 2025 • 0 new comments

• [Build Presets] Update public docs to use build presets

  #11559 commented on Aug 21, 2025 • 0 new comments

• [0.7 QA] Profiling Tutorial page is broken

  #12777 commented on Aug 21, 2025 • 0 new comments

• [Build Presets] Create a Windows preset

  #10723 commented on Aug 21, 2025 • 0 new comments

• [Build Presets] Create an Android preset

  #10720 commented on Aug 21, 2025 • 0 new comments

• [CMake] Enable BUILD_SHARED_LIBS=ON flag

  #10676 commented on Aug 21, 2025 • 0 new comments

• [Android] Add a Runtime.java

  #10439 commented on Aug 21, 2025 • 0 new comments

• [minibench] error report when run failed

  #10400 commented on Aug 21, 2025 • 0 new comments

• [torchao] finalize affine quant primitives

  #9782 commented on Aug 21, 2025 • 0 new comments

• Migrate ExecuTorch documentation to the new theme

  #10366 commented on Aug 21, 2025 • 0 new comments

• Handle 4-bit CoreML affine quantization via PT2E

  #9649 commented on Aug 21, 2025 • 0 new comments

• Support RuntimeSpec

  #9623 commented on Aug 21, 2025 • 0 new comments

• Support reshapeFrequency runtime spec

  #9619 commented on Aug 21, 2025 • 0 new comments

• Support exporting QNN models with Python wheels out-of-the-box

  #9474 commented on Aug 21, 2025 • 0 new comments

• [Android] Flatten data structure like dict in arg types

  #8905 commented on Aug 21, 2025 • 0 new comments

• Expose runtime error codes / results in JNI

  #8260 commented on Aug 21, 2025 • 0 new comments

• A supported ANE-friendly llama model with good out-of-the-box performance and accuracy.

  #9615 commented on Aug 21, 2025 • 0 new comments

• KernelCall failed at instruction 0:16 in operator aten::le.Tensor_out: 0x12

  #9433 commented on Aug 21, 2025 • 0 new comments

• Benchmark: Fix metrics high variability from run-to-run

  #8603 commented on Aug 21, 2025 • 0 new comments

• Build pip wheel and validate AOT workflow on different backends & host environments in CI

  #9019 commented on Aug 21, 2025 • 0 new comments

• Resolve the discrepancy of latency report between LLMs and non-LLMs

  #8576 commented on Aug 21, 2025 • 0 new comments

• Update Cadence Xtensa backend docs using backend template

  #8533 commented on Aug 21, 2025 • 0 new comments

• Update Qualcomm backend docs using backend template

  #8531 commented on Aug 21, 2025 • 0 new comments

• Update Vulkan backend docs using backend template

  #8529 commented on Aug 21, 2025 • 0 new comments

• Update MPS backend docs using backend template

  #8528 commented on Aug 21, 2025 • 0 new comments

• Benchmark Hugging Face models via Optimum integration

  #8294 commented on Aug 21, 2025 • 0 new comments