Repository Issue Activity (beta)

#23495 keras.ops.size returns int64 on four of five backends but declares int32

#23493 Add keras.ops.cov

#23499 Add `keras.ops.float_power`

#23497 Add `keras.ops.copysign`

#23487 Add label_smoothing to SparseCategoricalCrossentropy

#23479 Support for mixed_float8

#23485 Update all usages of `backend.is_tensor` to `backend.ops.is_tensor`

#23442 Feature request: Add SparseCategoricalFocalCrossentropy

#18957 Issue with Backpropagation on keras.ops.while_loop (in Jax)

#23370 I think Adam was chosen for the MNIST dataset because it combines the advantages of other optimizers and usually provides fast and reliable training.

#23451 Categorical_crossentropy and categorical_focal_crossentropy ignore axis for smoothing and warnings

#23452 Searchsorted in TF backend loses static shape info and performs redundant conversions

#23462 Torch backend: stateful GRU using cuDNN fails during backward due to an in-place state update

#23461 Feature Request: BDH layer

#23126 [Feature request] A Restoration and Perceptual loss suite

#20931 TF+XLA+Mixed Precision: Keras fails to compute gradients

#23459 pip install -r requirements.txt

#23460 Black Yung

#23453 [Fix] Resolve PyTorch DDP execution crashes, NCHW shape errors, and sampler state

#23400 keras.ops.numpy.i0 overflows to Inf for finite float64 result near x=713

#22835 [Bug] Input validation and logical inconsistency in MeanIoU segmentation metrics

#23441 test_eigh is silently skipped on the OpenVINO backend, and it fails when it runs

#19717 `plot_model` does not work for all models in `keras.applications`

#23244 [Feature Request] Auto-merge LoRA weights after training

#23413 [Bug] Division by zero in gptq_quantize_matrix