133 Pull requests merged by 41 people

• Add new export LLM config

  #11511 merged Jun 10, 2025

• Refactor FlatTensor

  #11499 merged Jun 10, 2025

• Use llm_config instead of args in export_llama functions

  #11162 merged Jun 10, 2025

• Convert args to LlmConfig

  #11081 merged Jun 10, 2025

• Introduce hydra framework with backwards compatibility

  #11029 merged Jun 10, 2025

• Add new export LLM config

  #11028 merged Jun 10, 2025

• Tests failing for exp op in xnnpack

  #11498 merged Jun 10, 2025

• [ET-VK][ez] Fix Vulkan Validation layer errors due to consecutive command buffer encoding

  #11488 merged Jun 10, 2025

• [ET-VK] Use shared pointer for vTensorStorage

  #11487 merged Jun 10, 2025

• [ET-VK][ez][testing] Remove conv2d test in compute_api_test + reduce stdout logging

  #11486 merged Jun 10, 2025

• Use nullptr to represent fallback kernels

  #11484 merged Jun 9, 2025

• GraphBuilder in memory passes unit tests. #3

  #11444 merged Jun 9, 2025

• Move gflags to third-party

  #11448 merged Jun 9, 2025

• Dont require method loading for meta meta in module api

  #11496 merged Jun 9, 2025

• fix unittest-buck2 -- buck test doesn't imply buck build

  #11388 merged Jun 9, 2025

• Adding operators: bitwiseand, bitwiseor, bitwisexor, embedding, eq, fmod, ge, gt, hardtanh, le, lt, maskedfill, ne, select_copy and quantisedmatmul. Link all kernels to internal compiler (#10840)

  #11402 merged Jun 9, 2025

• Migrate some missed pt2e pieces

  #11500 merged Jun 9, 2025

• Implement unary_ufunc functions using elementwise_util

  #9386 merged Jun 9, 2025

• Add mixed integer precision test for op_mul

  #11206 merged Jun 9, 2025

• Add SupportedTensorDtypes::BOOL

  #9584 merged Jun 9, 2025

• relax tolerances for all unary float ops

  #9585 merged Jun 9, 2025

• Add vectorization in elementwise_util

  #9432 merged Jun 9, 2025

• [ET-VK] Minor dispatch improvement to conv2d dw op to improve performance.

  #11495 merged Jun 9, 2025

• [ET-VK] Adding more test cases for conv 2d dw op.

  #11494 merged Jun 9, 2025

• Add optimized_portable_kernels test

  #11205 merged Jun 9, 2025

• Add vectorized_math.h

  #11204 merged Jun 9, 2025

• Migrate xnnpack/vulkan/boltnn pt2e from torch.ao to torchao

  #11363 merged Jun 9, 2025

• Remove setting BUCK2 when building the wheel

  #11490 merged Jun 9, 2025

• [ET-VK] Minor dispatch improvement to conv2d dw op to improve performance.

  #11477 merged Jun 9, 2025

• [ET-VK] Adding more test cases for conv 2d dw op.

  #11476 merged Jun 9, 2025

• Enforce required cmake flags as explicit requirements

  #11457 merged Jun 9, 2025

• [ET-VK][ez] Fix Vulkan Validation layer errors due to consecutive command buffer encoding

  #11401 merged Jun 9, 2025

• [ET-VK] Use shared pointer for vTensorStorage

  #11400 merged Jun 9, 2025

• [ET-VK][ez][testing] Remove conv2d test in compute_api_test + reduce stdout logging

  #11452 merged Jun 9, 2025

• NXP backend: Enable initial unit tests workflow

  #11396 merged Jun 9, 2025

• Arm backend: Make validate_same_dtype print readable types

  #11468 merged Jun 9, 2025

• Arm backend: Re-enable Dedicated_Sram for the Ethos-U85

  #11459 merged Jun 8, 2025

• adjust compatibility for the cadence backend

  #11439 merged Jun 7, 2025

• adjust compatibility for qualcomm Python backend

  #11438 merged Jun 7, 2025

• adjust compatibility for the cortex_m backend

  #11440 merged Jun 7, 2025

• Revert 11294 v2

  #11465 merged Jun 7, 2025

• Use GraphBuilder in test_replace_ops_passes. #2

  #11456 merged Jun 7, 2025

• Starter Task 1: Get learning rate for llm_pte_finetuning example from config file

  #11445 merged Jun 7, 2025

• Allow us to print output metadata for the case where output is not tensor

  #11263 merged Jun 7, 2025

• Revert "Implement GLU using internal views to avoid copying (#11295)"

  #11464 merged Jun 7, 2025

• Create a cmake function to test for conflicting flags

  #11460 merged Jun 6, 2025

• Create a function to check if dependent flags are ON

  #11455 merged Jun 6, 2025

• Fix-up

  #11443 merged Jun 6, 2025

• Resolve buck earlier

  #11451 merged Jun 6, 2025

• Use GraphBuilder in test_replace_ops_passes. #1

  #11344 merged Jun 6, 2025

• [pick][llm] Add a generic text only LLM runner

  #11441 merged Jun 6, 2025

• Complex Support: optimized bmm

  #11428 merged Jun 6, 2025

• Fix oss ci death test

  #11442 merged Jun 6, 2025

• Use GraphBuilder in memory passes unit tests. # 2

  #11292 merged Jun 6, 2025

• Integrate the IntermediateOutputCapturer into Inspector

  #11384 merged Jun 6, 2025

• [ET-VK] Adding more pointwise convolution tests.

  #11435 merged Jun 6, 2025

• Fix default value of cat_dim.

  #11403 merged Jun 6, 2025

• Arm backend: Support for dynamic shapes and fix resize bugs

  #11310 merged Jun 6, 2025

• [llm] Add a generic text only LLM runner

  #11342 merged Jun 6, 2025

• fix model export issues

  #11397 merged Jun 6, 2025

• [ET-VK] Adding more pointwise convolution tests.

  #11358 merged Jun 6, 2025

• Init Value with a generic scalar.

  #11429 merged Jun 6, 2025

• Qualcomm AI Engine Direct - GA MobileVit V2

  #11279 merged Jun 6, 2025

• Update Cortex-M Size target

  #11399 merged Jun 5, 2025

• Milestone3.1: Support tanh op in XNNPACK backend

  #11364 merged Jun 5, 2025

• Qualcomm AI Engine Direct - GA mobilevit

  #11208 merged Jun 5, 2025

• Qualcomm AI Engine Direct - Streaming Mimi Enablement

  #10570 merged Jun 5, 2025

• Qualcomm AI Engine Direct - GA CvT

  #11036 merged Jun 5, 2025

• Qualcomm AI Engine Direct - QAIRT Visualizer Engagement

  #10873 merged Jun 5, 2025

• Neuron buffer allocator decouple from ExecuTorch framework

  #8760 merged Jun 5, 2025

• Reduce memory requirement on export_llama tests with no params (#11175)

  #11339 merged Jun 5, 2025

• Generate 1 kernel libary with dtype selective build and used it in multiple executorch_generated_lib

  #11378 merged Jun 5, 2025

• Add index check for embedding kernel

  #11375 merged Jun 5, 2025

• Arm backend: Replace asserts with exceptions in passes

  #11394 merged Jun 5, 2025

• Add visibility to codegen's gen_lib so users can use it

  #11386 merged Jun 5, 2025

• Introduce PAL function table

  #10675 merged Jun 5, 2025

• Qualcomm AI Engine Direct - GA DIT

  #11093 merged Jun 5, 2025

• Qualcomm AI Engine Direct - GA FocalNet

  #11097 merged Jun 5, 2025

• Restore explicit so load in Module.java

  #11385 merged Jun 5, 2025

• fix binary_ops buck deps

  #11379 merged Jun 5, 2025

• Temp fix to unblock diff train

  #11361 merged Jun 5, 2025

• Create decompose_ops.py and test_decompose_ops.py

  #11299 merged Jun 5, 2025

• remove raied error to solve ci issue.

  #11376 merged Jun 5, 2025

• [ET-VK] Support setting cache_data_path at build time

  #11383 merged Jun 5, 2025

• Use __cpp_aligned_new instead of hand-rolling the implementation

  #11293 merged Jun 4, 2025

• Support exp op in XNNPACK backend

  #11373 merged Jun 4, 2025

• Revert "Switch fbcode builds of ExecuTorch and PyTorch to fbsource sl…

  #11374 merged Jun 4, 2025

• Script for benchmark stability assessment

  #10982 merged Jun 4, 2025

• [ET-VK] Use push constants for image and buffer to nchw prepack nodes.

  #11371 merged Jun 4, 2025

• [ET-VK] Support setting cache_data_path at run time

  #11382 merged Jun 4, 2025

• [ET-VK] Consolidate shader compilation into one vkCreateComputePipelines call

  #11381 merged Jun 4, 2025

• [ET-VK][Ops] aten.var.dim from scratch implementation

  #11380 merged Jun 4, 2025

• fix-up

  #11362 merged Jun 4, 2025

• [ET-VK] Support setting cache_data_path at build time

  #11359 merged Jun 4, 2025

• [ET-VK] Support setting cache_data_path at run time

  #11350 merged Jun 4, 2025

• [ET-VK] Consolidate shader compilation into one vkCreateComputePipelines call

  #11345 merged Jun 4, 2025

• [ET-VK][Ops] aten.var.dim from scratch implementation

  #11197 merged Jun 4, 2025

• [ET-VK] Using push constants for convolution prepack dispatches.

  #11372 merged Jun 4, 2025

• Make exec_aten::optional and string_view direct aliases for std versions

  #11321 merged Jun 4, 2025

• support channels last inputs in xnnpack

  #11118 merged Jun 4, 2025

• Implement GLU using internal views to avoid copying

  #11295 merged Jun 4, 2025

• Arm backend: Avoid subtraction error on boolean tensors in error diff logging

  #11249 merged Jun 4, 2025

• [ET-VK] Using push constants for convolution prepack dispatches.

  #11306 merged Jun 4, 2025

• [ET-VK] Use push constants for image and buffer to nchw prepack nodes.

  #11305 merged Jun 4, 2025

• Added docs for setting up executorch on windows

  #11273 merged Jun 4, 2025

• check for overflow in calculate_nbytes

  #11217 merged Jun 4, 2025

• Support more open-source models

  #11125 merged Jun 4, 2025

• Make op_glu_test input asymmetric

  #11294 merged Jun 4, 2025

• Switch fbcode builds of ExecuTorch and PyTorch to fbsource sleef

  #11352 merged Jun 4, 2025

• Save 16 KiB by moving part of optimized add/sub to .cpp

  #11346 merged Jun 4, 2025

• Arm backend: Replace asserts with exceptions in quantization annotator

  #11356 merged Jun 4, 2025

• Arm backend: Remove special host build of flatcc

  #11278 merged Jun 4, 2025

• Bump PyTorch pin to dev20250602

  #11318 merged Jun 4, 2025

• [ET-VK] Migrate ops to use DynamicDispatchNode

  #11353 merged Jun 4, 2025

• [ET-VK] Migrate ops to use DynamicDispatchNode

  #11312 merged Jun 4, 2025

• bump torchao pin

  #11340 merged Jun 4, 2025

• Switch fbcode builds of ExecuTorch and PyTorch to fbsource sleef

  #11261 merged Jun 4, 2025

• irangeify most of runtime/core/exec_aten

  #8612 merged Jun 4, 2025

• [ET-VK][ez] Fix handling of assert ops

  #11349 merged Jun 4, 2025

• [ET-VK][Ops] aten.tan.default from scratch implementation

  #11152 merged Jun 4, 2025

• Dtype selective build: disable when not in xplat and add fail msg

  #11219 merged Jun 4, 2025

• [ET-VK] Add support for binary symint ops

  #11348 merged Jun 4, 2025

• [ET-VK] Add reshape functions for transformers related operators

  #11347 merged Jun 4, 2025

• Reusable test framework the Inspector tests

  #11314 merged Jun 4, 2025

• [ET-VK][ez] Fix handling of assert ops

  #11258 merged Jun 4, 2025

• [ET-VK] Add support for binary symint ops

  #11257 merged Jun 4, 2025

• [ET-VK] Add reshape functions for transformers related operators

  #11256 merged Jun 4, 2025

• Removing uneeded buck deps in cmake build flow

  #11315 merged Jun 3, 2025

• Mitigate coremltool issue with python 3.13

  #11317 merged Jun 3, 2025

• Consolidating executor runners in CMake - xnnpack

  #11239 merged Jun 3, 2025

• [ET-VK] Using push constants for buffer to image prepack nodes.

  #11308 merged Jun 3, 2025

• Arm Backend: Remove unused param in aot_arm_compiler.py

  #10970 merged Jun 3, 2025

• Bump tokenizers

  #11300 merged Jun 3, 2025

59 Pull requests opened by 33 people

• [Draft] Qualcomm AI Engine Direct - Unexpected graph for mutable buffer after export during Quantization

  #11309 opened Jun 3, 2025

• Milestone2.2: Optimize transposes in XNNPACK partition by removing redundant to_copy ops

  #11316 opened Jun 3, 2025

• Replace fixed number cmake --build -j in sh and docs #10887

  #11351 opened Jun 4, 2025

• Qualcomm AI Engine Direct - GA Model Enablement (Roberta)

  #11354 opened Jun 4, 2025

• Qualcomm AI Engine Direct - CI for Non-LLM GA model Part1

  #11357 opened Jun 4, 2025

• [ET-VK] double, short, and uint16 dtype runtime support

  #11365 opened Jun 4, 2025

• [ET-VK][Ops] quantize ops skeleton test framework

  #11366 opened Jun 4, 2025

• [ET-VK][Ops] quantize_per_token.default test setup

  #11367 opened Jun 4, 2025

• [ET-VK][Ops] quantize_per_tensor.default test setup

  #11368 opened Jun 4, 2025

• [ET-VK][Ops] quantization op shaders and impl

  #11369 opened Jun 4, 2025

• [1/N] Add BackendOptions class

  #11389 opened Jun 5, 2025

• [2/N] Add update context

  #11390 opened Jun 5, 2025

• [3/N] Add update function in backend interface

  #11391 opened Jun 5, 2025

• Qualcomm AI Engine Direct - Deprecate convert_bmm_to_matmul pass

  #11392 opened Jun 5, 2025

• Ability to set external instance + devices

  #11393 opened Jun 5, 2025

• Use Memcpy in copy_utils

  #11430 opened Jun 5, 2025

• Plumbing for dtype selection in cmake flow - header generation

  #11432 opened Jun 5, 2025

• Update tutorial-arm-ethos-u.md

  #11433 opened Jun 5, 2025

• Qualcomm AI Engine Direct - multi-method support in to_edge_transform_and_lower_to_qnn

  #11436 opened Jun 6, 2025

• Qualcomm AI Engine Direct - Enable Lookahead Decoding

  #11437 opened Jun 6, 2025

• Fix typo in native_batch_norm

  #11446 opened Jun 6, 2025

• Benchmark HF optimum-executorch

  #11450 opened Jun 6, 2025

• Updated list of supported chipsets for Qualcomm AI Engine backend

  #11453 opened Jun 6, 2025

• Gather intermediate output from Events

  #11454 opened Jun 6, 2025

• Automatically determine presets for known platforms

  #11461 opened Jun 6, 2025

• [4/N] Add backend options map

  #11462 opened Jun 6, 2025

• [5/N] Add update in method

  #11463 opened Jun 6, 2025

• Try some QD8-BF16 Experiments

  #11466 opened Jun 7, 2025

• Update documents for buffer allocator

  #11467 opened Jun 9, 2025

• Allow passing `install_executorch.py` args to `backends/openvino/scripts/openvino_build.sh`

  #11469 opened Jun 9, 2025

• Elaborate on the release notes labels for backends

  #11470 opened Jun 9, 2025

• Arm backend: Add validate_valid_dtype

  #11472 opened Jun 9, 2025

• NXP backend: Add docker image for nxp backend tests

  #11473 opened Jun 9, 2025

• Arm backend: Add skip setup options to setup.sh

  #11474 opened Jun 9, 2025

• Arm backend: Introduce TOSA aware context

  #11475 opened Jun 9, 2025

• [ET] enabling half dtype input for quantization

  #11479 opened Jun 9, 2025

• [ET-VK][Ops] dequantize ops skeleton test framework

  #11480 opened Jun 9, 2025

• [ET-VK][Ops] dequantize_per_tensor.default test setup

  #11481 opened Jun 9, 2025

• [ET-VK][Ops] dequantize_per_token.default test setup

  #11482 opened Jun 9, 2025

• [ET-VK][Ops] dequantization op shaders and impl

  #11483 opened Jun 9, 2025

• Avoid use-after-return caused by double move

  #11485 opened Jun 9, 2025

• Typecheck 25% of exir directory

  #11489 opened Jun 9, 2025

• FlatTensor refactor (combined)

  #11491 opened Jun 9, 2025

• Add legacy mode test

  #11492 opened Jun 9, 2025

• Configure conflicting ARM baremetal cmake flags

  #11493 opened Jun 9, 2025

• Make core aten exception ops modifiable at runtime.

  #11497 opened Jun 9, 2025

• Use GraphBuilder in test_replace_ops_passes. #3

  #11501 opened Jun 9, 2025

• Typecheck 42% of exir directory

  #11502 opened Jun 9, 2025

• bump nightly pin

  #11503 opened Jun 9, 2025

• Extracting dtype information from model PTE in OSS

  #11504 opened Jun 9, 2025

• [xnn update prep] Update Dtypes for XNNPACK

  #11505 opened Jun 9, 2025

• [xnn update prep] deprecate sdpa

  #11506 opened Jun 9, 2025

• [xnn update prep] Update ET Pthreadpool third-party to google

  #11507 opened Jun 9, 2025

• [ET-VK] New implementation of `cat` operator

  #11508 opened Jun 9, 2025

• Reapply #11294 and #11295 (improve GLU test and implement using internal views to avoid copying)

  #11509 opened Jun 10, 2025

• Bump tokenizers

  #11510 opened Jun 10, 2025

• Introduce hydra framework with backwards compatibility

  #11512 opened Jun 10, 2025

• Convert args to LlmConfig

  #11513 opened Jun 10, 2025

• Use llm_config instead of args in export_llama functions

  #11514 opened Jun 10, 2025

18 Issues closed by 12 people

• Building ExecuTorch on RPi5 with Clang 14.0.6 fails due to bfloat incompatibility

  #8924 closed Jun 9, 2025

• ./install_executorch.sh doesn't work on Mac OS in latest main branch [42c4cdba]

  #11478 closed Jun 9, 2025

• Add complex dtype support and tests for: sum.dim_IntList

  #11423 closed Jun 5, 2025

• Add complex dtype support and tests for: mul.Tensor

  #11411 closed Jun 5, 2025

• Add complex dtype support and tests for: bmm

  #11425 closed Jun 5, 2025

• Undefined symbol while compiling runtime for MediaTek

  #10389 closed Jun 5, 2025

• mediatek backend.so build failed

  #7549 closed Jun 5, 2025

• Unsafe prebuilt boundary in MediaTek backend

  #6074 closed Jun 5, 2025

• Add complex dtype support and tests for: diagonal_copy

  #11422 closed Jun 5, 2025

• Add complex dtype support and tests for: view_as_real_copy

  #11406 closed Jun 5, 2025

• Provide an alternative PAL override mechanism

  #11142 closed Jun 5, 2025

• PT2E Dynamic Quantization with XNNPACK Backend for Android fails Loading of method forward in runtime

  #11355 closed Jun 5, 2025

• TorchAudio Conformer model doesn't lower to edge IR

  #7998 closed Jun 5, 2025

• [MediaTek] mtk_neuron compile error: 3D permutation not support

  #9733 closed Jun 5, 2025

• Clean up ET CoreML documentation

  #9624 closed Jun 4, 2025

• Dynamic shape support

  #9618 closed Jun 4, 2025

• Don't Serialize Scales/ZP in Flatbuffer

  #9029 closed Jun 3, 2025

• Mechanism to detect performance CPU core is not reliable

  #10549 closed Jun 3, 2025

48 Issues opened by 12 people

• Export fails: sdpa_with_kv_cache not registered despite clean build from source (_portable_lib missing)

  #11471 opened Jun 9, 2025

• Clean up turned on flags for pybind preset

  #11458 opened Jun 6, 2025

• Use fuse/expand_dims for squeeze unsqueeze

  #11449 opened Jun 6, 2025

• Pass Dependencies when Proposing Partitions

  #11447 opened Jun 6, 2025

• Enable selective build model API in OSS

  #11434 opened Jun 5, 2025

• Unable to run install_executorch.sh on macOS

  #11427 opened Jun 5, 2025

• Add complex dtype support and tests for: _fft_c2r

  #11426 opened Jun 5, 2025

• Add complex dtype support and tests for: _linalg_solve_ex

  #11424 opened Jun 5, 2025

• Add complex dtype support and tests for: repeat

  #11421 opened Jun 5, 2025

• Add complex dtype support and tests for: view_copy

  #11420 opened Jun 5, 2025

• Add complex dtype support and tests for: where.self

  #11419 opened Jun 5, 2025

• Add complex dtype support and tests for: full

  #11418 opened Jun 5, 2025

• Add complex dtype support and tests for: div.Tensor

  #11417 opened Jun 5, 2025

• Add complex dtype support and tests for: cumsum

  #11416 opened Jun 5, 2025

• Add complex dtype support and tests for: permute_copy

  #11415 opened Jun 5, 2025

• Add complex dtype support and tests for: clone

  #11414 opened Jun 5, 2025

• Add complex dtype support and tests for: _conj_copy

  #11413 opened Jun 5, 2025

• Add complex dtype support and tests for: add.Tensor

  #11412 opened Jun 5, 2025

• Add complex dtype support and tests for: _to_dim_order_copy

  #11410 opened Jun 5, 2025

• Add complex dtype support and tests for: unsqueeze_copy

  #11409 opened Jun 5, 2025

• Add complex dtype support and tests for: squeeze_copy.dims

  #11408 opened Jun 5, 2025

• Add complex dtype support and tests for: index_select

  #11407 opened Jun 5, 2025

• Add complex dtype support and tests for: _fft_r2c

  #11405 opened Jun 5, 2025

• Portable Library: Complex Support

  #11404 opened Jun 5, 2025

• Draft RFC on extended error context

  #11398 opened Jun 5, 2025

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

  #11395 opened Jun 5, 2025

• Import and extend c10/util/safe_numerics.h

  #11370 opened Jun 4, 2025

• x86_64 emulator Tests failing

  #11341 opened Jun 3, 2025

• Implement portable kernel: native_layer_norm_backward.out

  #11337 opened Jun 3, 2025

• Implement portable kernel: native_group_norm_backward.out

  #11336 opened Jun 3, 2025

• Implement portable kernel: embedding_dense_backward.out

  #11335 opened Jun 3, 2025

• Implement portable kernel: avg_pool2d_backward.grad_input

  #11334 opened Jun 3, 2025

• Implement portable kernel: _adaptive_avg_pool2d_backward.out

  #11333 opened Jun 3, 2025

• Implement portable kernel: _fft_r2c.out

  #11332 opened Jun 3, 2025

• Implement portable kernel: sort.values

  #11331 opened Jun 3, 2025

• Implement portable kernel: scatter_reduce.two_out

  #11330 opened Jun 3, 2025

• Implement portable kernel: max_pool3d_with_indices.out

  #11329 opened Jun 3, 2025

• Implement portable kernel: grid_sampler_2d.out

  #11328 opened Jun 3, 2025

• Implement portable kernel: col2im.out

  #11327 opened Jun 3, 2025

• Implement portable kernel: avg_pool3d.out

  #11326 opened Jun 3, 2025

• Implement portable kernel: avg_pool1d.out

  #11325 opened Jun 3, 2025

• Implement portable kernel: adaptive_avg_pool1d.out

  #11324 opened Jun 3, 2025

• Implement portable kernel: _embedding_bag.out

  #11323 opened Jun 3, 2025

• Implement portable kernel: _adaptive_avg_pool3d.out

  #11322 opened Jun 3, 2025

• Portable Library: Core-ATen op coverage

  #11320 opened Jun 3, 2025

• Implement portable kernel: _adaptive_avg_pool2d.out

  #11319 opened Jun 3, 2025

• High-level api for canned user experience on devtool

  #11313 opened Jun 3, 2025

• Running Qwen3 in Android LlamaDemo App shows an error while loading tokenizer

  #11311 opened Jun 3, 2025

37 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.

• [Example] Yolo12 Detection sample with OpenVINO/XNNPACK backend

  #10156 commented on Jun 6, 2025 • 10 new comments

• [RFC] Fetch gflags dynamically

  #11285 commented on Jun 9, 2025 • 2 new comments

• NXP backend: Add support for depthwise and separable convolution.

  #11215 commented on Jun 3, 2025 • 2 new comments

• Add et_dump related API to module

  #9284 commented on Jun 9, 2025 • 1 new comment

• Qualcomm AI Engine Direct - Enable custom operator

  #8726 commented on Jun 9, 2025 • 1 new comment

• [ExecuTorch][#10447] Extend `PyBundledModule` with `extension.BundledModule`

  #10450 commented on Jun 9, 2025 • 0 new comments

• Use std::string_view and std::optional

  #10541 commented on Jun 3, 2025 • 0 new comments

• Introducing NXP Neutron runtime

  #10563 commented on Jun 3, 2025 • 0 new comments

• Clean up eager quant in llm_export

  #10684 commented on Jun 10, 2025 • 0 new comments

• Add aot example with Neutron Backend

  #10871 commented on Jun 3, 2025 • 0 new comments

• [pytorch hash update] update the pinned pytorch hash

  #10955 commented on Jun 10, 2025 • 0 new comments

• [ET-VK][Ops] aten.tan.default in unary_ops

  #11153 commented on Jun 4, 2025 • 0 new comments

• Enforce clang for Windows builds

  #11159 commented on Jun 3, 2025 • 0 new comments

• adjust platform compatibility for qnn backend

  #11179 commented on Jun 9, 2025 • 0 new comments

• NOCOMMIT: Bisect CoreML failures when bumping PyTorch pin

  #11222 commented on Jun 3, 2025 • 0 new comments

• Remove outdated editable mode warning from building from source docs

  #11245 commented on Jun 4, 2025 • 0 new comments

• fix and link

  #11267 commented on Jun 4, 2025 • 0 new comments

• Improve torch.ops typing

  #11276 commented on Jun 4, 2025 • 0 new comments

• Milestone2.1: Partition to_dim_order_copy op in XNN delegate

  #11286 commented on Jun 5, 2025 • 0 new comments

• [WIP] Device telemetry for benchmark

  #11301 commented on Jun 4, 2025 • 0 new comments

• Complie issue cause lost #include

  #11307 commented on Jun 3, 2025 • 0 new comments

• Get rid of fixed number cmake --build -j in sh and docs

  #10887 commented on Jun 4, 2025 • 0 new comments

• Manual kernel registration to include library names in API

  #11221 commented on Jun 5, 2025 • 0 new comments

• Optimized add fails fails at runtime with rank error when broadcasting

  #10959 commented on Jun 6, 2025 • 0 new comments

• Issue with run the Cadence demo on Ubuntu 22.04

  #11050 commented on Jun 6, 2025 • 0 new comments

• torchao build fails on macOS arm64 due to missing symbol c10::StorageImpl::throw_data_ptr_access_error()

  #11283 commented on Jun 6, 2025 • 0 new comments

• which clang source of bf16 GEMM kernel

  #10513 commented on Jun 9, 2025 • 0 new comments

• Remove ExecuTorch copy of Vectorized

  #7042 commented on Jun 10, 2025 • 0 new comments

• Add MediaTek backend CI

  #7120 commented on Jun 9, 2025 • 0 new comments

• fix spec_prop_pass

  #7974 commented on Jun 6, 2025 • 0 new comments

• Fix comment in memory_planning.py

  #8010 commented on Jun 6, 2025 • 0 new comments

• remove the exec_aten namespace

  #8018 commented on Jun 6, 2025 • 0 new comments

• Revert to use mean_out than mean_dim_out

  #8021 commented on Jun 6, 2025 • 0 new comments

• [docs][ez] Fix doc build workflow

  #8079 commented on Jun 6, 2025 • 0 new comments

• Adjust tolerance for quantized XNN conv1d tests

  #8093 commented on Jun 6, 2025 • 0 new comments

• [devtool] create stream_data_sink

  #8604 commented on Jun 6, 2025 • 0 new comments

• Arm backend: Enable mypy lintrunner for backends/arm/test

  #9357 commented on Jun 4, 2025 • 0 new comments