pytorch/executorch

View on GitHub

On-device AI across mobile, embedded and edge for PyTorch

Stars: 5k
Forks: 1.2k
Open beginner issues: 0
Indexed issues: 851
Avg merge: 2d 10h
Merged PRs (30d): 581

Dominant language: Python
License: No license data
Last GitHub push: Sep 19, 2026
Latest indexed: Sep 19, 2026
Contributing guide: Contributing guide
Code of conduct: Code of conduct
Beginner labels: good first issue help wanted

Beginner friendly Everything

Open All

851 open issues indexed

crash while using the android xnn runner Open

module: xnnpack

pytorch/executorch#16241 · 8 comments · 2 assignees · Opened Dec 14, 2025
Need a tool to analyze a pte file and list all required backends and ops Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

pytorch/executorch#16237 · 1 comment · 1 reaction · Opened Dec 13, 2025
etLLM changing model export based on how it's named is very confusing Open

pytorch/executorch#16203 · 1 comment · 1 assignee · Opened Dec 11, 2025
Update XNNPACK dependency Open

module: xnnpack triaged

pytorch/executorch#16201 · 1 comment · 1 assignee · Opened Dec 11, 2025
[Windows] Installation fails with "No CUDA toolset found" when CUDA 13.0 is installed Open

Difficulty 4/5 3-5 days Newbie friendliness 38/100

pytorch/executorch#16191 · 2 comments · Opened Dec 11, 2025
Demo apps should not hardcode templates/model types Open

pytorch/executorch#16190 · 1 comment · 1 assignee · Opened Dec 11, 2025
Evaluate memory bound op performance on CPU Open

pytorch/executorch#16158 · 1 assignee · Opened Dec 9, 2025
Cross attention KV cache support in optimum-executorch Open

pytorch/executorch#16157 · 1 assignee · Opened Dec 9, 2025
Sample logits on CUDA memory before copying out to CPU Open

pytorch/executorch#16156 · 1 assignee · Opened Dec 9, 2025
Is dynamic weight update / fine-tuning supported in QNN / XNNPACK backends? Open

module: training

Difficulty 5/5 Over a week Newbie friendliness 25/100

pytorch/executorch#16123 · 5 comments · Opened Dec 7, 2025
LSTM: Add CMSIS Op Support Open

Difficulty 5/5 Over a week Newbie friendliness 15/100

pytorch/executorch#16108 · Opened Dec 5, 2025
DecomposeGroupNormPass(ArmPass) fails Open

module: arm partner: arm

Difficulty 4/5 3-5 days Newbie friendliness 38/100

pytorch/executorch#16090 · 3 comments · Opened Dec 3, 2025
Scope / prototype YNNPACK in ET XNN backend Open

pytorch/executorch#16046 · 1 assignee · Opened Dec 2, 2025
CORTEX_M: Add Packed int4 support Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

pytorch/executorch#16043 · Opened Dec 2, 2025
CORTEX_M: Add Int16 operator support Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

pytorch/executorch#16042 · Opened Dec 2, 2025
CORTEX_M: Quantization configuration Open

pytorch/executorch#16040 · 1 assignee · Opened Dec 2, 2025
CORTEX_M: Target configuration Open

pytorch/executorch#16039 · 1 assignee · Opened Dec 2, 2025
How to add a new backend? Open

module: backend

pytorch/executorch#16034 · 4 comments · 1 assignee · Opened Dec 2, 2025
exporting qwen3-0.6B using the w8a8 quantization format, a segmentation error occurs. Open

module: qnn partner: qualcomm

pytorch/executorch#16013 · 15 comments · 1 assignee · Opened Dec 1, 2025
How to run add operator in executorch ? Open

module: runtime

pytorch/executorch#16010 · 3 comments · 1 assignee · Opened Nov 30, 2025
qnn_executor_runner - mismatch in the skel files ? Open

module: qnn partner: qualcomm

Difficulty 4/5 3-5 days Newbie friendliness 28/100

pytorch/executorch#15978 · 3 comments · Opened Nov 25, 2025
What should I do if there is SoC for my processor? Open

module: qnn partner: qualcomm

Difficulty 4/5 3-5 days Newbie friendliness 25/100

pytorch/executorch#15973 · 3 comments · Opened Nov 25, 2025
Update doc links for 1.0.1 Open

pytorch/executorch#15966 · 1 assignee · Opened Nov 24, 2025
CUDA backend benchmark ci Open

pytorch/executorch#15921 · 1 assignee · Opened Nov 20, 2025
Add segments for bpte Open

pytorch/executorch#15874 · 2 comments · 1 reaction · 2 assignees · Opened Nov 18, 2025
Pybindings for flat tensor Open

pytorch/executorch#15873 · 1 comment · 1 assignee · Opened Nov 18, 2025
is executorch distribute for Thor and Spark? Open

Difficulty 5/5 Over a week Newbie friendliness 15/100

pytorch/executorch#15825 · 4 comments · Opened Nov 14, 2025
Why force_fp32 is set to true for conv weight and biases on XNNPack backend Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

pytorch/executorch#15817 · 3 comments · 1 reaction · Opened Nov 13, 2025
Upgrade pytest to v9 Open

module: ci

Difficulty 4/5 3-5 days Newbie friendliness 35/100

pytorch/executorch#15797 · 1 reaction · Opened Nov 12, 2025
support custom triton sdpa kernel in CUDA backend Open

pytorch/executorch#15744 · 2 comments · 1 assignee · Opened Nov 11, 2025
Performance comparison between whisper.cpp and ET with CUDA & Vulkan backends Open

module: vulkan

pytorch/executorch#15742 · 1 assignee · Opened Nov 11, 2025
QNN export failure for PARSeq text recognition model Open

module: qnn partner: qualcomm

pytorch/executorch#15734 · 15 comments · 1 reaction · 1 assignee · Opened Nov 11, 2025
please support normal java and scala3 Open

module: android

Difficulty 5/5 Over a week Newbie friendliness 18/100

pytorch/executorch#15688 · 25 comments · Opened Nov 8, 2025
Which Backend is supported for Google Pixel 9 and 10? Open

Difficulty 4/5 3-5 days Newbie friendliness 20/100

pytorch/executorch#15670 · 2 comments · Opened Nov 7, 2025
Serializing int32_t dtypes Open

Difficulty 5/5 Over a week Newbie friendliness 38/100

pytorch/executorch#15669 · 1 comment · 1 reaction · Opened Nov 7, 2025
Enable Sleef in CMake builds Open

Difficulty 3/5 1-2 days Newbie friendliness 42/100

pytorch/executorch#15658 · Opened Nov 6, 2025
Performance comparison between whisper.cpp and ET cuda backend, vulkan backend Open

module: vulkan

pytorch/executorch#15642 · 1 assignee · Opened Nov 6, 2025
Use FQNs instead of hash for tensor/data names Open

pytorch/executorch#15560 · 1 assignee · Opened Nov 4, 2025
Exposing intermediate tensor allocations as constants or graph inputs Open

pytorch/executorch#15505 · 6 comments · 1 assignee · Opened Oct 31, 2025
Invalid result of running U2Net on Vulkan Open

module: vulkan

pytorch/executorch#15490 · 10 comments · 1 assignee · Opened Oct 31, 2025
Fuzzer test alignment Open

pytorch/executorch#15466 · 1 assignee · Opened Oct 30, 2025
Consolidate metal & cuda backend in AOT and runtime Open

pytorch/executorch#15424 · 2 comments · 3 assignees · Opened Oct 28, 2025
3P integration planning Open

pytorch/executorch#15423 · 1 assignee · Opened Oct 28, 2025
qnn_llama_runner on SA8295 outputs repetitive “)” or gibberish with Qwen3-0.6B after ExecuTorch export Open

module: qnn partner: qualcomm

Difficulty 4/5 3-5 days Newbie friendliness 28/100

pytorch/executorch#15410 · 11 comments · Opened Oct 28, 2025
Refactor CI job HF setup Open

pytorch/executorch#15404 · 1 assignee · Opened Oct 27, 2025
Export a forward-only PTE model (with forward() returns loss + mutable params) for gradient-free finetuning Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

pytorch/executorch#15389 · 2 comments · Opened Oct 26, 2025
Executorch with QNN AI Engine Backend Open

module: qnn partner: qualcomm

Difficulty 4/5 3-5 days Newbie friendliness 25/100

pytorch/executorch#15387 · 26 comments · Opened Oct 25, 2025
Vulkan backend inference is non-deterministic with stabilityai/sd-vae-ft-mse exported via ExecuTorch (no quantization used) Open

module: vulkan

pytorch/executorch#15344 · 1 comment · 1 assignee · Opened Oct 22, 2025
Error in minimal example code for export with vulken as backend Open

pytorch/executorch#15296 · 2 comments · 1 assignee · Opened Oct 21, 2025
Model support: IBM Granite-4.0-H-Micro Open

pytorch/executorch#15248 · 2 comments · 1 assignee · Opened Oct 18, 2025

pytorch/executorch

Get new issues in your inbox