Repository Issue Activity (beta)

#8894 xdsl dependancy requirement (ai-edge-litert[model-utils])

#8913 GPU delegate ignores `BATCH_MATMUL` `adj_x` and `adj_y` options

#8657 NpuAcceleratorProvider.isDeviceSupported() returns false for SM8450 despite presence of AOT compiled models

#8906 ml_dtypes should be a base dependency, not gated behind the model-utils extra (ai-edge-litert 2.1.6)

#8903 Importing tensor from `AHardwareBuffer` fails on Adreno 660

#8889 Builtin no-delegate interpreter corrupts reused buffer under default memory planner (RESHAPE→RMSNorm region)

#8608 BATCH_MATMUL: heap OOB read on a rank<2 operand

#8609 Interpreter: heap OOB read on an out-of-range node input tensor index

#8610 LSTM / UNIDIRECTIONAL_SEQUENCE_LSTM: heap OOB read of weights rank in Prepare

#8831 Google-Tensor AOT: blockwise-quantized model fails to serialize — "Failed to find scales tensor in map during serialization"

#8869 [Qualcomm NPU] Compiler plugin does not repartition after QNN VTCM allocation failure (Gemma 4 E2B, SM8845)

#8884 [Google Tensor SDK Beta][Tensor G5] AOT compiler returns INTERNAL for a fully partitioned 3.48 GB multimodal TFLite model

#8835 Git-LFS problem - Release v2.1.6 artifacts contain fake binary content of 133 bytes in size

#8474 LiteRT 2.1.6 upgrade failed

#7598 [Bug] AOT compilation succeeds for MediaPipe selfie model but fails for YOLO11n on Tensor G4 — INTERNAL error at compiler_plugin.cc:481

#8787 LiteRtMetalAccelerator cannot be registered on iphone

#7713 [Feature Request] Native support for 1-bit and Ternary (1.58-bit) quantization

#8521 SPM checkout fails: LFS smudge error

#8098 LiteRT MediaTek compiler plugin fails on per-channel quantized Conv2D int32 bias with Neuron SDK 8.2.19

#8223 [AOT] Qualcomm SocModel enum missing SM8475 (Snapdragon 8+ Gen 1), already in supported_soc.csv

#8446 Qwen3-0.6B Google Tensor AOT failed

#8567 [TFLite] FULLY_CONNECTED reference kernel should perform requantization using integer kernels instead of float

#6745 iOS arm64 libLiteRtMetalAccelerator.dylib missing/mispackaged in litert_prebuilts.zip

#8586 [feature windows]: Better D3D interop

#4421 Package still includes 'backports-strenum' even if python version 3.12+