Weekly issue metrics report - 2025-03-01..2025-03-07

[github-actions]

Issue Metrics

Metric	Average	Median	90th percentile
Time to first response	14:18:01	4:14:53	21:27:39
Time to close	1 day, 20:29:55	1 day, 10:53:52	3 days, 14:08:38
Time to answer	None	None	None
Time spent in module: android	6 days, 11:38:42	6 days, 11:38:42	6 days, 13:22:58
Time spent in module: ios	None	None	None
Time spent in module: benchmark	None	None	None
Time spent in module: build/install	4 days, 7:10:29	3 days, 17:16:19	7 days, 9:24:18
Time spent in module: vulkan	None	None	None
Time spent in module: llm	4 days, 15:54:16	4 days, 2:40:14	7 days, 15:10:35
Time spent in module: ci	None	None	None
Time spent in module: training	8:40:54	8:40:54	8:40:54
Time spent in module: user experience	None	None	None

Metric	Count
Number of items that remain open	44
Number of items closed	6
Number of most active mentors	0
Total number of items created	50

Title	URL	Author	Time to first response	Time to close	Time to answer	Time spent in module: android	Time spent in module: ios	Time spent in module: benchmark	Time spent in module: build/install	Time spent in module: vulkan	Time spent in module: llm	Time spent in module: ci	Time spent in module: training	Time spent in module: user experience
Dynamic Parameters for Mediatek Delegate	pytorch/executorch#9035	shreshth-tru	11:05:25	None	None	None	None	None	None	None	None	None	None	None
[Request Help] I refer the Delegate Debugging document to debug the QLoRA_LLama 3.2-1B model, and want to get the pipeline of linear operation delegation and determine the underlying kernel in XNNPACK	pytorch/executorch#9034	liye0626	14:03:51	None	None	None	None	None	None	None	None	None	None	None
[Request Help] I would like to ask, does ExecuTorch have plans to support EAGLE-1/EAGLE-2/EAGLE-3?	pytorch/executorch#9032	liye0626	0:36:22	23:39:28	None	None	None	None	None	None	23:02:45	None	None	None
Don't Serialize Scales/ZP in Flatbuffer	pytorch/executorch#9029	mcr229	2:43:30	None	None	None	None	None	None	None	None	None	None	None
Refactor binary op partitioner configs under binary op config class	pytorch/executorch#9024	mcr229	None	None	None	None	None	None	None	None	None	None	None	None
Partitioner to not delegate mixed dtype operators	pytorch/executorch#9023	mcr229	None	None	None	None	None	None	None	None	None	None	None	None
Duplicate Dynamic Quant Chain in XNNPACK Lowering Flow	pytorch/executorch#9022	mcr229	None	None	None	None	None	None	None	None	None	None	None	None
Support Dynamically Quantized Convolutions	pytorch/executorch#9021	mcr229	None	None	None	None	None	None	None	None	None	None	None	None
Build pip wheel and validate AOT workflow on different backends & host environments in CI	pytorch/executorch#9019	larryliu0820	None	None	None	None	None	None	3 days, 7:52:23.703432	None	None	None	None	None
SDPA custom op works without kv cach	pytorch/executorch#9014	iseeyuan	None	None	None	None	None	None	None	None	3 days, 7:54:22.981972	None	None	None
Freeing unpacked weights immediately after packing	pytorch/executorch#9013	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
Shared activation memory with ExecuTorch	pytorch/executorch#9012	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
Setup Build-Size Check in CI for XNNPACK Dependency	pytorch/executorch#9011	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
Binary Size: Selective Build	pytorch/executorch#9010	cbilgin	None	None	None	None	None	None	3 days, 7:50:55.055117	None	None	None	None	None
Increased Op Coverage:	pytorch/executorch#9009	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
Dynamism Support in XNNPACK for vision models	pytorch/executorch#9008	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
Model Load Time: SIMD Packing Kernels for Blockwise Quantization	pytorch/executorch#9007	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
Inference Latency + Memory Usage: Proper Dim Order Support	pytorch/executorch#9006	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
Hardened FP16 Support	pytorch/executorch#9005	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
BF16 Support	pytorch/executorch#9004	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
Memory Usage: XNNPACK Weights Cache + Blob Storage Service	pytorch/executorch#9003	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
Model File Size: Backend Weight Sharing via. Blob Storage Service	pytorch/executorch#9002	cbilgin	None	None	None	None	None	None	None	None	None	None	None	None
Quantized MobileBert Model Fails with XNNPACK Assertion Error on Android Deployment	pytorch/executorch#8994	qxuan512	None	None	None	None	None	None	None	None	None	None	None	None
Training pybindings can't be installed	pytorch/executorch#8990	YuanTingHsieh	13:33:06	22:14:11	None	None	None	None	8:40:31	None	None	None	8:40:54	None
[Request Help] “torch._dynamo.exc.UserError: Could not guard on data-dependent expression Eq(u77 - 1, 1)"	pytorch/executorch#8988	liye0626	None	None	None	None	None	None	None	None	None	None	None	None
Enable dynamic shapes for torchao:8daXw quant ops	pytorch/executorch#8981	jackzhxng	None	None	None	None	None	None	None	None	None	None	None	None
Editable mode error for mac installation	pytorch/executorch#8980	jackzhxng	1 day, 18:40:42	1 day, 22:08:16	None	None	None	None	1 day, 22:08:10	None	None	None	None	None
Fail to export QNN model on SXR2230P SOC	pytorch/executorch#8973	nirvanagth	1:16:12	None	None	None	None	None	None	None	None	None	None	None
Reduce the number of files / directories at the root of the repository	pytorch/executorch#8966	byjlw	None	None	None	None	None	None	None	None	None	None	None	None
Replace remaining uses of linearize_access_indexes for broadcasting with BroadcastIndexesRange	pytorch/executorch#8965	swolchok	None	None	None	None	None	None	None	None	None	None	None	None
[XNNPACK Backend] Modified MobileBERT model outputs original shape on Android despite successful desktop inference	pytorch/executorch#8956	qxuan512	None	2:40:21	None	None	None	None	None	None	None	None	None	None
Enable ExecuTorch Tensor to store arbitrary resources	pytorch/executorch#8939	SS-JIA	None	None	None	None	None	None	None	None	None	None	None	None
toorchao: quantization schemes can't find ao lib	pytorch/executorch#8937	jackzhxng	None	None	None	None	None	None	None	None	None	None	None	None
Program data separation delegate support	pytorch/executorch#8935	lucylq	None	None	None	None	None	None	None	None	None	None	None	None
Parallelize portable ops if threadpool is available, with fallback to parallel_for-as-for-loop	pytorch/executorch#8932	swolchok	0:23:41	None	None	None	None	None	None	None	None	None	None	None
Building ExecuTorch on RPi5 with Clang 14.0.6 fails due to bfloat incompatibility	pytorch/executorch#8924	spalatinate	0:39:58	None	None	None	None	None	5 days, 11:13:15.954491	None	None	None	None	None
Error running .pte model with executor_runner	pytorch/executorch#8923	corehalt	4:05:13	None	None	None	None	None	None	None	None	None	None	None
Memory Profiling in DevTools	pytorch/executorch#8911	GregoryComer	0:29:07	None	None	None	None	None	None	None	None	None	None	None
[Android] Flatten data structure like dict in arg types	pytorch/executorch#8905	kirklandsign	None	None	None	6 days, 9:28:21.224879	None	None	None	None	None	None	None	None
Enable bias quantization for prequantized checkpoints	pytorch/executorch#8903	jackzhxng	None	None	None	None	None	None	None	None	None	None	None	None
Missing operator: [3] cadence::quantized_relu.per_tensor_out when following build-run-xtensa tutorial	pytorch/executorch#8900	ELR11C	0:51:05	2 days, 21:24:19	None	None	None	None	None	None	None	None	None	None
XNNPack configurations can be complicated to get right	pytorch/executorch#8884	tarun292	1:30:32	None	None	None	None	None	None	None	None	None	None	None
Building llava cpp runner fails	pytorch/executorch#8881	spalatinate	4:14:53	4 days, 6:52:57	None	None	None	None	4 days, 2:40:14	None	4 days, 2:40:14	None	None	None
Temperature Settings in Android llama demo (Mediatek)	pytorch/executorch#8878	shreshth-tru	9:15:54	None	None	6 days, 13:49:02.384731	None	None	None	None	6 days, 13:49:02.384740	None	None	None
Building wheel for executorch (pyproject.toml) did not run successfully.	pytorch/executorch#8874	ManojShanmugamTruminds	10:27:46	None	None	None	None	None	6 days, 13:46:57.915577	None	None	None	None	None
[Request Help] “torch._dynamo.exc.UserError"	pytorch/executorch#8873	liye0626	16:09:23	None	None	None	None	None	None	None	None	None	None	None
Using Hugging Face model card name in export_llama	pytorch/executorch#8872	iseeyuan	None	None	None	None	None	None	None	None	8 days, 8:04:56.716104	None	None	None
Successfully registered all kernels from shared library: NOT_SUPPORTED	pytorch/executorch#8870	ismukhin	5 days, 7:35:39	None	None	None	None	None	None	None	None	None	None	None
Failure to set up Executorch	pytorch/executorch#8869	adonnini	9:59:56	None	None	None	None	None	None	None	None	None	None	None
Add upstream flatcc option to avoid installing under the source tree	pytorch/executorch#8867	dbort	None	None	None	None	None	None	9 days, 7:11:25.888017	None	None	None	None	None

This report was generated with the Issue Metrics Action
Search query used to find these items: repo:pytorch/executorch is:issue created:2025-03-01..2025-03-07