Search results

userApi/tensor: promote requantizeTensorInPlace from example code to TensorApi when a second consumer appears

Status: Open.

#387 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 20, 2026

arithmetic/executeOp: writeOut swap-restore of the target's qConfig roundingMode is not reentrant — replace with explicit rounding parameter when multi-threading lands

Status: Open.

#386 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 20, 2026

tensor: ASYM int-repr dequant codes[i] + zeroPoint can overflow int32 for far-from-zero positive grids

Status: Open.

#384 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 20, 2026

optimizer: LR warmup for lrScheduler_t (linear warmup, composable with existing schedulers)

enhancementNew feature or requestNew feature or request

Status: Open.

#383 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 20, 2026

optimizer: global-norm gradient clipping (clip_grad_norm_ parity)

enhancementNew feature or requestNew feature or request

Status: Open.

#382 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 20, 2026

data_loader: per-epoch reshuffle — DataLoader shuffles once at init, diverging from PyTorch semantics

enhancementNew feature or requestNew feature or request

Status: Open.

#381 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 20, 2026

optimizer/training_loop: layer freezing & partial-update policy — per-layer trainable flag with backward truncation (epic)

enhancementNew feature or requestNew feature or request

Status: Open.

#380 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 20, 2026

arithmetic: enable TRACK_INSTRUCTIONS on the legacy op libs

Status: Open.

#351 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 13, 2026

training-state checkpointing: optimizer/scheduler/RNG state is not serializable — resume silently corrupts training

enhancementNew feature or requestNew feature or request

Status: Open.

#350 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 13, 2026

build: FP contraction diverges host vs MCU — decide global -ffp-contract policy

enhancementNew feature or requestNew feature or request

Status: Open.

#349 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 13, 2026

continual-learning: post-#326 hardening follow-ups (test gaps, CCIPCA in-place, CI gate order)

enhancementNew feature or requestNew feature or request

Status: Open.

#346 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 13, 2026

data_loader: transform/targetTransform fields are declared but never invoked

bugSomething isn't workingSomething isn't working

Status: Open.

#345 In es-ude/OnDeviceTraining;

· LeoBuron opened on Jul 13, 2026