Issues · tensorflow/tensorflow · GitHub

Labels Milestones New issue

`tf.linalg.det` gradient crashes on singular matrices

#119477

· wuyii8941 opened

on May 26, 2026

`tf.math.xlogy` and `tf.math.xlog1py` gradient w.r.t. x is incorrectly zero when x=0

#119476

· wuyii8941 opened

on May 26, 2026

TFLite mnist_tflite tutorial is not runnable as written

#119439

· fallintoplace opened

on May 25, 2026

XLA `tf.nn.top_k` returns different results from eager execution on NaN input

#119438

· MRiffiAslett opened

on May 25, 2026

`tf.math.cumprod` gradient is incorrect when input contains zeros

#119430

· wuyii8941 opened

on May 25, 2026

Grappler ArithmeticOptimizer incorrectly eliminates `reciprocal(reciprocal(x))` as involution

#119429

· wuyii8941 opened

on May 25, 2026

Grappler constant folding optimizes `x * 0` → `0`, suppressing NaN for Inf/NaN inputs

#119428

· wuyii8941 opened

on May 25, 2026

[TFLite] FULLY_CONNECTED reference kernel should use MultiplyByQuantizedMultiplier instead of std::round for requantization

#119412

· KavinSai-synthara opened

on May 25, 2026

`tf.quantization.fake_quant_with_min_max_vars/ _per_channel` silent value mismatch under XLA compilation breaks QAT

#119411

· rookieLiu2018 opened

on May 25, 2026

Internal BroadcastGradientArgs shape error when differentiating tf.boolean_mask followed by weighted reduction

#119382

· rookieLiu2018 opened

on May 24, 2026

`tf.unique` (and `boolean_mask`, `dynamic_partition`) silently produces corrupt output for 16-bit dtypes on GPU XLA

#119380

· wuyii8941 opened

on May 24, 2026

TensorArray(dynamic_size=True, size=0) inside while_loop silently discards writes

awaiting PR merge

#119375

· rookieLiu2018 opened

on May 24, 2026